Follow Discord
Sweep 25 Sep 2026 · 19:33Z Build v2.1.283 504 read Stable v2.1.274 Latest v2.1.283 Next v2.1.283 Feeds RSS JSON llms.txt llms-full.txt Unofficial
One capture · api

One read of Claude Developer Platformapi-20260925T153714Z

310 pages moved out of 637 read.

Pages moved 310 significant first
Pages read 637 in this capture
Captured 15:37 UTC
Corpus hash 850f99a0d461 corpus-hash

What this read moved

126-150 of 310, page 6 of 13

This capture is too large to show at once. Changes 126-150 of 310 are below, significant first; the rest are on the following screens.

api/compliance/apps/sessions/remote Changed · +10 / -10 lines

from line 59
5959 
6060 Maximum results (default: 100, max: 500)
6161 
62 default: 100, maximum: 500, minimum: 1
62 default: 100, minimum: 1, maximum: 500
6363 
6464- `organization_ids: optional array of string`
6565 
from line 113
113113 
114114 - `started_by_user: object or null`
115115 
116 A user associated with a remote session.
116 The user who initiated an agent-owned session (for example, by mentioning Claude in Slack or via a scheduled trigger). Null for user-owned sessions — where the session's `user` started it — and for agent sessions with no human initiator. For initiators no longer a member of an organization the key may read, the object is populated with `email_address` null.
117117 
118118 - `id: string`
119119 
from line 135
135135 
136136 - `user: object or null`
137137 
138 A user associated with a remote session.
138 The user who owns the session. Null for sessions owned by an automated agent rather than a user. At most one of `user` and `agent_id` is set. For users no longer a member of an organization the key may read, the object is populated with `email_address` null.
139139 
140140 - `id: string`
141141 
from line 220
220220 
221221 - `started_by_user: object or null`
222222 
223 A user associated with a remote session.
223 The user who initiated an agent-owned session (for example, by mentioning Claude in Slack or via a scheduled trigger). Null for user-owned sessions — where the session's `user` started it — and for agent sessions with no human initiator. For initiators no longer a member of an organization the key may read, the object is populated with `email_address` null.
224224 
225225 - `id: string`
226226 
from line 242
242242 
243243 - `user: object or null`
244244 
245 A user associated with a remote session.
245 The user who owns the session. Null for sessions owned by an automated agent rather than a user. At most one of `user` and `agent_id` is set. For users no longer a member of an organization the key may read, the object is populated with `email_address` null.
246246 
247247 - `id: string`
248248 
from line 291
291291 
292292 Maximum results (default: 100, max: 1000)
293293 
294 default: 100, maximum: 1000, minimum: 1
294 default: 100, minimum: 1, maximum: 1000
295295 
296296- `order: optional "asc" or "desc"`
297297 
from line 311
311311 
312312 Truncate each text item inside a tool result to at most this many bytes (cut on a code-point boundary). Pass `-1` to request the server maximum. `0` is not a valid value.
313313 
314 default: 10000, maximum: 2147483647, minimum: -1
314 default: 10000, minimum: -1, maximum: 2147483647
315315 
316316- `tool_use_input_max_bytes: optional number`
317317 
318318 Truncate each tool-use input to at most this many bytes (cut on a code-point boundary so the result is valid UTF-8). Pass `-1` to request the server maximum. `0` is not a valid value.
319319 
320 default: 10000, maximum: 2147483647, minimum: -1
320 default: 10000, minimum: -1, maximum: 2147483647
321321 
322322#### Headers
323323 
from line 479
479479 
480480 - `started_by_user: object or null`
481481 
482 A user associated with a remote session.
482 The user who initiated an agent-owned session (for example, by mentioning Claude in Slack or via a scheduled trigger). Null for user-owned sessions — where the session's `user` started it — and for agent sessions with no human initiator. For initiators no longer a member of an organization the key may read, the object is populated with `email_address` null.
483483 
484484 - `id: string`
485485 
from line 501
501501 
502502 - `user: object or null`
503503 
504 A user associated with a remote session.
504 The user who owns the session. Null for sessions owned by an automated agent rather than a user. At most one of `user` and `agent_id` is set. For users no longer a member of an organization the key may read, the object is populated with `email_address` null.
505505 
506506 - `id: string`
507507 

api/compliance/apps/sessions/remote/list Changed · +3 / -3 lines

from line 57
5757 
5858 Maximum results (default: 100, max: 500)
5959 
60 default: 100, maximum: 500, minimum: 1
60 default: 100, minimum: 1, maximum: 500
6161 
6262- `organization_ids: optional array of string`
6363 
from line 111
111111 
112112 - `started_by_user: object or null`
113113 
114 A user associated with a remote session.
114 The user who initiated an agent-owned session (for example, by mentioning Claude in Slack or via a scheduled trigger). Null for user-owned sessions — where the session's `user` started it — and for agent sessions with no human initiator. For initiators no longer a member of an organization the key may read, the object is populated with `email_address` null.
115115 
116116 - `id: string`
117117 
from line 133
133133 
134134 - `user: object or null`
135135 
136 A user associated with a remote session.
136 The user who owns the session. Null for sessions owned by an automated agent rather than a user. At most one of `user` and `agent_id` is set. For users no longer a member of an organization the key may read, the object is populated with `email_address` null.
137137 
138138 - `id: string`
139139 

api/compliance/apps/sessions/remote/messages Changed · +5 / -5 lines

from line 42
4242 
4343 Maximum results (default: 100, max: 1000)
4444 
45 default: 100, maximum: 1000, minimum: 1
45 default: 100, minimum: 1, maximum: 1000
4646 
4747- `order: optional "asc" or "desc"`
4848 
from line 62
6262 
6363 Truncate each text item inside a tool result to at most this many bytes (cut on a code-point boundary). Pass `-1` to request the server maximum. `0` is not a valid value.
6464 
65 default: 10000, maximum: 2147483647, minimum: -1
65 default: 10000, minimum: -1, maximum: 2147483647
6666 
6767- `tool_use_input_max_bytes: optional number`
6868 
6969 Truncate each tool-use input to at most this many bytes (cut on a code-point boundary so the result is valid UTF-8). Pass `-1` to request the server maximum. `0` is not a valid value.
7070 
71 default: 10000, maximum: 2147483647, minimum: -1
71 default: 10000, minimum: -1, maximum: 2147483647
7272 
7373### Headers
7474 
from line 230
230230 
231231 - `started_by_user: object or null`
232232 
233 A user associated with a remote session.
233 The user who initiated an agent-owned session (for example, by mentioning Claude in Slack or via a scheduled trigger). Null for user-owned sessions — where the session's `user` started it — and for agent sessions with no human initiator. For initiators no longer a member of an organization the key may read, the object is populated with `email_address` null.
234234 
235235 - `id: string`
236236 
from line 252
252252 
253253 - `user: object or null`
254254 
255 A user associated with a remote session.
255 The user who owns the session. Null for sessions owned by an automated agent rather than a user. At most one of `user` and `agent_id` is set. For users no longer a member of an organization the key may read, the object is populated with `email_address` null.
256256 
257257 - `id: string`
258258 

api/compliance/apps/sessions/remote/messages/list Changed · +5 / -5 lines

from line 40
4040 
4141 Maximum results (default: 100, max: 1000)
4242 
43 default: 100, maximum: 1000, minimum: 1
43 default: 100, minimum: 1, maximum: 1000
4444 
4545- `order: optional "asc" or "desc"`
4646 
from line 60
6060 
6161 Truncate each text item inside a tool result to at most this many bytes (cut on a code-point boundary). Pass `-1` to request the server maximum. `0` is not a valid value.
6262 
63 default: 10000, maximum: 2147483647, minimum: -1
63 default: 10000, minimum: -1, maximum: 2147483647
6464 
6565- `tool_use_input_max_bytes: optional number`
6666 
6767 Truncate each tool-use input to at most this many bytes (cut on a code-point boundary so the result is valid UTF-8). Pass `-1` to request the server maximum. `0` is not a valid value.
6868 
69 default: 10000, maximum: 2147483647, minimum: -1
69 default: 10000, minimum: -1, maximum: 2147483647
7070 
7171## Headers
7272 
from line 228
228228 
229229 - `started_by_user: object or null`
230230 
231 A user associated with a remote session.
231 The user who initiated an agent-owned session (for example, by mentioning Claude in Slack or via a scheduled trigger). Null for user-owned sessions — where the session's `user` started it — and for agent sessions with no human initiator. For initiators no longer a member of an organization the key may read, the object is populated with `email_address` null.
232232 
233233 - `id: string`
234234 
from line 250
250250 
251251 - `user: object or null`
252252 
253 A user associated with a remote session.
253 The user who owns the session. Null for sessions owned by an automated agent rather than a user. At most one of `user` and `agent_id` is set. For users no longer a member of an organization the key may read, the object is populated with `email_address` null.
254254 
255255 - `id: string`
256256 

api/compliance/code Changed · +2 / -7 lines

from line 30
3030 
3131 Maximum results (default: 20, max: 100)
3232 
33 default: 20, maximum: 100, minimum: 1
33 default: 20, minimum: 1, maximum: 100
3434 
3535- `organization_ids: optional array of string`
3636 
from line 120
120120 
121121 - `user: object or null`
122122 
123 The user who owns a Code Artifact.
124 
125 Fields that reference this type are null when the Artifact was
126 published by an agent session rather than a user account, when the
127 owner's account has been deleted, or when the owner is no longer a
128 member of an organization the key may read.
123 Artifact owner with email, or null if the Artifact was published by an agent session, the owner's account has been deleted, or the owner is no longer a member of an organization the key may read
129124 
130125 - `id: string`
131126 

api/compliance/code/artifacts Changed · +3 / -13 lines

from line 28
2828 
2929 Maximum results (default: 20, max: 100)
3030 
31 default: 20, maximum: 100, minimum: 1
31 default: 20, minimum: 1, maximum: 100
3232 
3333- `organization_ids: optional array of string`
3434 
from line 118
118118 
119119 - `user: object or null`
120120 
121 The user who owns a Code Artifact.
121 Artifact owner with email, or null if the Artifact was published by an agent session, the owner's account has been deleted, or the owner is no longer a member of an organization the key may read
122122 
123 Fields that reference this type are null when the Artifact was
124 published by an agent session rather than a user account, when the
125 owner's account has been deleted, or when the owner is no longer a
126 member of an organization the key may read.
127 
128123 - `id: string`
129124 
130125 User identifier (tagged ID)
from line 328
333328 
334329 - `user: object or null`
335330 
336 The user who owns a Code Artifact.
337 
338 Fields that reference this type are null when the Artifact was
339 published by an agent session rather than a user account, when the
340 owner's account has been deleted, or when the owner is no longer a
341 member of an organization the key may read.
331 Artifact owner with email, or null if the Artifact was published by an agent session, the owner's account has been deleted, or the owner is no longer a member of an organization the key may read
342332 
343333 - `id: string`
344334 

api/compliance/code/artifacts/list Changed · +2 / -7 lines

from line 26
2626 
2727 Maximum results (default: 20, max: 100)
2828 
29 default: 20, maximum: 100, minimum: 1
29 default: 20, minimum: 1, maximum: 100
3030 
3131- `organization_ids: optional array of string`
3232 
from line 116
116116 
117117 - `user: object or null`
118118 
119 The user who owns a Code Artifact.
120 
121 Fields that reference this type are null when the Artifact was
122 published by an agent session rather than a user account, when the
123 owner's account has been deleted, or when the owner is no longer a
124 member of an organization the key may read.
119 Artifact owner with email, or null if the Artifact was published by an agent session, the owner's account has been deleted, or the owner is no longer a member of an organization the key may read
125120 
126121 - `id: string`
127122 

api/compliance/organizations Changed · +4 / -4 lines

from line 21
2121 
2222 Maximum results (default: 1000, max: 1000)
2323 
24 default: 1000, maximum: 1000, minimum: 1
24 default: 1000, minimum: 1, maximum: 1000
2525 
2626- `page: optional string`
2727 
from line 121
121121 
122122 Maximum results (default: 500, max: 1000)
123123 
124 default: 500, maximum: 1000, minimum: 1
124 default: 500, minimum: 1, maximum: 1000
125125 
126126- `page: optional string`
127127 
from line 235
235235 
236236 Maximum results (default: 500, max: 1000)
237237 
238 default: 500, maximum: 1000, minimum: 1
238 default: 500, minimum: 1, maximum: 1000
239239 
240240- `page: optional string`
241241 
from line 399
399399 
400400 Maximum results (default: 500, max: 1000)
401401 
402 default: 500, maximum: 1000, minimum: 1
402 default: 500, minimum: 1, maximum: 1000
403403 
404404- `page: optional string`
405405 

api/files Changed · +9 / -9 lines

from line 59
5959 
6060 Original filename of the uploaded file.
6161 
62 maxLength: 500, minLength: 1
62 minLength: 1, maxLength: 500
6363 
6464 - `mime_type: string`
6565 
6666 MIME type of the file.
6767 
68 maxLength: 255, minLength: 1
68 minLength: 1, maxLength: 255
6969 
7070 - `size_bytes: number`
7171 
from line 128
128128 
129129 Defaults to `20`. Ranges from `1` to `1000`.
130130 
131 default: 20, maximum: 1000, minimum: 1
131 default: 20, minimum: 1, maximum: 1000
132132 
133133- `page: optional string`
134134 
from line 170
170170 
171171 Original filename of the uploaded file.
172172 
173 maxLength: 500, minLength: 1
173 minLength: 1, maxLength: 500
174174 
175175 - `mime_type: string`
176176 
177177 MIME type of the file.
178178 
179 maxLength: 255, minLength: 1
179 minLength: 1, maxLength: 255
180180 
181181 - `size_bytes: number`
182182 
from line 302
302302 
303303 Original filename of the uploaded file.
304304 
305 maxLength: 500, minLength: 1
305 minLength: 1, maxLength: 500
306306 
307307 - `mime_type: string`
308308 
309309 MIME type of the file.
310310 
311 maxLength: 255, minLength: 1
311 minLength: 1, maxLength: 255
312312 
313313 - `size_bytes: number`
314314 
from line 449
449449 
450450 Original filename of the uploaded file.
451451 
452 maxLength: 500, minLength: 1
452 minLength: 1, maxLength: 500
453453 
454454 - `mime_type: string`
455455 
456456 MIME type of the file.
457457 
458 maxLength: 255, minLength: 1
458 minLength: 1, maxLength: 255
459459 
460460 - `size_bytes: number`
461461 

api/files/list Changed · +3 / -3 lines

from line 21
2121 
2222 Defaults to `20`. Ranges from `1` to `1000`.
2323 
24 default: 20, maximum: 1000, minimum: 1
24 default: 20, minimum: 1, maximum: 1000
2525 
2626- `page: optional string`
2727 
from line 63
6363 
6464 Original filename of the uploaded file.
6565 
66 maxLength: 500, minLength: 1
66 minLength: 1, maxLength: 500
6767 
6868 - `mime_type: string`
6969 
7070 MIME type of the file.
7171 
72 maxLength: 255, minLength: 1
72 minLength: 1, maxLength: 255
7373 
7474 - `size_bytes: number`
7575 

api/messages Changed · +856 / -987 lines

This page is larger than the 256 KiB this site keeps, so one side of the diff below stops where the stored text does.

from line 141
141141 
142142 - `document_title: string or null`
143143 
144 maxLength: 500, minLength: 1
144 minLength: 1, maxLength: 500
145145 
146146 - `end_char_index: number`
147147 
from line 161
161161 
162162 - `document_title: string or null`
163163 
164 maxLength: 500, minLength: 1
164 minLength: 1, maxLength: 500
165165 
166166 - `end_page_number: number`
167167 
from line 185
185185 
186186 - `document_title: string or null`
187187 
188 maxLength: 500, minLength: 1
188 minLength: 1, maxLength: 500
189189 
190190 - `end_block_index: number`
191191 
from line 209
209209 
210210 - `title: string or null`
211211 
212 maxLength: 512, minLength: 1
212 minLength: 1, maxLength: 512
213213 
214214 - `url: string`
215215 
from line 365
365365 
366366 - `title: optional string or null`
367367 
368 maxLength: 500, minLength: 1
368 minLength: 1, maxLength: 500
369369 
370370 - `SearchResultBlockParam object`
371371 
from line 429
429429 
430430 - `name: string`
431431 
432 maxLength: 200, minLength: 1
432 minLength: 1, maxLength: 200
433433 
434434 - `cache_control: optional CacheControlEphemeral or null`
435435 
from line 465
465465 
466466 For a toolset member tool_use, the toolset family this member belongs to.
467467 
468 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
468 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
469469 
470470 - `ToolResultBlockParam object`
471471 
from line 501
501501 
502502 - `tool_name: string`
503503 
504 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
504 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
505505 
506506 - `cache_control: optional CacheControlEphemeral or null`
507507 
from line 529
529529 
530530 The caller-assigned identifier for this tab, unique within the inventory.
531531 
532 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
532 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
533533 
534534 - `title: string`
535535 
from line 555
555555 
556556 Tabs opened and download state changes during this call. "Nothing to report" is expressed by omitting the field, never by an empty list.
557557 
558 maxItems: 200, minItems: 1
558 minItems: 1, maxItems: 200
559559 
560560 - `BrowserStateChangeTabOpened object`
561561 
from line 573
573573 
574574 The `tab_id` of the opened tab, present in `tabs`.
575575 
576 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
576 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
577577 
578578 - `BrowserStateChangeDownloadStarted object`
579579 
from line 585
585585 
586586 The caller-assigned identifier for this download, stable across the state changes reporting it.
587587 
588 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
588 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
589589 
590590 - `url: string`
591591 
from line 606
606606 
607607 The caller-assigned identifier for this download, stable across the state changes reporting it.
608608 
609 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
609 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
610610 
611611 - `url: string`
612612 
from line 618
618618 
619619 Where the executor saved the file, on the executor's filesystem. Only included when another tool in the same environment can read the file at that path.
620620 
621 pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$, maxLength: 4096
621 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
622622 
623623 - `size_bytes: optional number or null`
624624 
from line 636
636636 
637637 The caller-assigned identifier for this download, stable across the state changes reporting it.
638638 
639 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
639 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
640640 
641641 - `url: string`
642642 
from line 648
648648 
649649 The failure or cancellation detail, when known.
650650 
651 pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$, maxLength: 4096
651 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
652652 
653653 - `is_error: optional boolean`
654654 
from line 656
656656 
657657 For a toolset member tool_result, the toolset family of the paired tool_use.
658658 
659 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
659 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
660660 
661661 - `ServerToolUseBlockParam object`
662662 
from line 1034
10341034 
10351035 - `tool_name: string`
10361036 
1037 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
1037 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
10381038 
10391039 - `cache_control: optional CacheControlEphemeral or null`
10401040 
from line 1075
10751075 
10761076 See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.
10771077 
1078 - `string`
1079 
10781080 - `"claude-fable-5-1" or "claude-opus-5-5" or "claude-mythos-5-1" or 15 more`
10791081 
10801082 The model that will complete your prompt.
from line 1119
11171119 
11181120 Powerful intelligence for long-running agents and coding
11191121 
1120 - `"claude-mythos-preview"`
1121 
1122 New class of intelligence, strongest in coding and cybersecurity
1123 
11241122 - `"claude-opus-4-6"`
11251123 
11261124 Powerful intelligence for long-running agents and coding
from line 1151
11531151 
11541152 High-performance model for agents and coding
11551153 
1156 - `string`
1154 - `"claude-mythos-preview"`
11571155 
1156 **Deprecated**: Will reach end-of-life on June 30, 2026. Please migrate to claude-mythos-5. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.
1157 
1158 New class of intelligence, strongest in coding and cybersecurity
1159 
11581160- `cache_control: optional CacheControlEphemeral or null`
11591161 
11601162 Top-level cache control automatically applies a cache_control marker to the last cacheable block in the request.
from line 1191
11891191 
11901192 Skill ID
11911193 
1192 maxLength: 64, minLength: 1
1194 minLength: 1, maxLength: 64
11931195 
11941196 - `version: optional string`
11951197 
11961198 Skill version or 'latest' for most recent version
11971199 
1198 maxLength: 64, minLength: 1
1200 minLength: 1, maxLength: 64
11991201 
12001202 - `string`
12011203 
12021204- `diagnostics: optional DiagnosticsParam or null`
12031205 
1204 Request-level diagnostics. Currently carries the previous response
1205 id for prompt-cache divergence reporting.
1206 Request-level diagnostics. Supply `previous_message_id` to have the response include `diagnostics.cache_miss_reason` explaining any prompt-cache divergence from that prior request.
12061207 
12071208 - `previous_message_id: optional string or null`
12081209 
from line 1233
12321233 
12331234 - `effort: optional "low" or "medium" or "high" or 2 more or null`
12341235 
1235 All possible effort levels.
1236 How much effort the model should put into its response. Higher effort levels may result in more thorough analysis but take longer.
12361237 
1238 Valid values are `low`, `medium`, `high`, `xhigh`, or `max`.
1239 
12371240 - `"low"`
12381241 
12391242 - `"medium"`
from line 1485
14821485 
14831486 This is how the tool will be called by the model and in `tool_use` blocks.
14841487 
1485 maxLength: 128, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,128}$
1488 minLength: 1, maxLength: 128, pattern: ^[a-zA-Z0-9_-]{1,128}$
14861489 
14871490 - `allowed_callers: optional array of "direct" or "code_execution_20250825" or "code_execution_20260120" or "code_execution_20260521"`
14881491 
from line 1702
16991702 
17001703 - `configs: optional BrowserToolsetConfigs or null`
17011704 
1702 Per-member configuration for `browser_toolset_20260801`: one
1703 optional field per member tool, keyed by the member name — the same
1704 name the member's `tool_use` blocks carry. Every member is an
1705 accepted key, and a member's defaults apply wherever its key is
1706 absent. Unknown keys are rejected: the field set is this toolset
1707 version's complete member set.
1705 Sparse per-member overrides, keyed by member name. Absent, null, and {} are equivalent; a member's defaults apply wherever its key is absent.
17081706 
17091707 - `type: optional BrowserTypeConfig or null`
17101708 
from line 2129
21312129 
21322130 - `configs: optional ComputerToolsetConfigs or null`
21332131 
2134 Per-member configuration for `computer_toolset_20260801`: one
2135 optional field per member tool, keyed by the member name — the same
2136 name the member's `tool_use` blocks carry. Every member is an
2137 accepted key, and a member's defaults apply wherever its key is
2138 absent. Unknown keys are rejected: the field set is this toolset
2139 version's complete member set.
2132 Sparse per-member overrides, keyed by member name. Absent, null, and {} are equivalent; a member's defaults apply wherever its key is absent.
21402133 
21412134 - `type: optional ComputerTypeConfig or null`
21422135 
from line 2483
24902483 
24912484 Maximum number of times the tool can be used in the API request.
24922485 
2493 exclusiveMinimum: 0
2486 minimum: 1
24942487 
24952488 - `strict: optional boolean`
24962489 
from line 2499
25062499 
25072500 The city of the user.
25082501 
2509 maxLength: 255, minLength: 1
2502 minLength: 1, maxLength: 255
25102503 
25112504 - `country: optional string or null`
25122505 
25132506 The two letter [ISO country code](https://en.wikipedia.org/wiki/ISO_3166-1_alpha-2) of the user.
25142507 
2515 maxLength: 2, minLength: 2
2508 minLength: 2, maxLength: 2
25162509 
25172510 - `region: optional string or null`
25182511 
25192512 The region of the user.
25202513 
2521 maxLength: 255, minLength: 1
2514 minLength: 1, maxLength: 255
25222515 
25232516 - `timezone: optional string or null`
25242517 
25252518 The [IANA timezone](https://nodatime.org/TimeZones) of the user.
25262519 
2527 maxLength: 255, minLength: 1
2520 minLength: 1, maxLength: 255
25282521 
25292522 - `WebFetchTool20250910 object`
25302523 
from line 2563
25702563 
25712564 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
25722565 
2573 exclusiveMinimum: 0
2566 minimum: 1
25742567 
25752568 - `max_uses: optional number or null`
25762569 
25772570 Maximum number of times the tool can be used in the API request.
25782571 
2579 exclusiveMinimum: 0
2572 minimum: 1
25802573 
25812574 - `strict: optional boolean`
25822575 
from line 2577
25842577 
25852578 - `url_sources: optional WebFetchURLSources or null`
25862579 
2587 Which sources contribute to the set of URLs web fetch may fetch.
2580 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
25882581 
2589 Each key is a tagged variant: `user_input` is `all` or `none`; the
2590 two tool filters are `all`, `none`, `only` (only the named tools'
2591 results) or `except` (every result but the named tools'). A named tool
2592 must be declared in this request's `tools[]`.
2593 
25942582 - `client_tool_results: optional WebFetchURLSourceAll or WebFetchURLSourceNone or WebFetchURLSourceOnly or WebFetchURLSourceExcept`
25952583 
25962584 Which client tools' results contribute fetchable URLs: "all", "none", or an only or except list of client tool names from tools[].
from line 2701
27132701 
27142702 Maximum number of times the tool can be used in the API request.
27152703 
2716 exclusiveMinimum: 0
2704 minimum: 1
27172705 
27182706 - `strict: optional boolean`
27192707 
from line 2755
27672755 
27682756 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
27692757 
2770 exclusiveMinimum: 0
2758 minimum: 1
27712759 
27722760 - `max_uses: optional number or null`
27732761 
27742762 Maximum number of times the tool can be used in the API request.
27752763 
2776 exclusiveMinimum: 0
2764 minimum: 1
27772765 
27782766 - `strict: optional boolean`
27792767 
from line 2769
27812769 
27822770 - `url_sources: optional WebFetchURLSources or null`
27832771 
2784 Which sources contribute to the set of URLs web fetch may fetch.
2772 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
27852773 
2786 Each key is a tagged variant: `user_input` is `all` or `none`; the
2787 two tool filters are `all`, `none`, `only` (only the named tools'
2788 results) or `except` (every result but the named tools'). A named tool
2789 must be declared in this request's `tools[]`.
2790 
27912774 - `WebFetchTool20260309 object`
27922775 
27932776 Web fetch tool with use_cache parameter for bypassing cached content.
from line 2817
28342817 
28352818 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
28362819 
2837 exclusiveMinimum: 0
2820 minimum: 1
28382821 
28392822 - `max_uses: optional number or null`
28402823 
28412824 Maximum number of times the tool can be used in the API request.
28422825 
2843 exclusiveMinimum: 0
2826 minimum: 1
28442827 
28452828 - `strict: optional boolean`
28462829 
from line 2831
28482831 
28492832 - `url_sources: optional WebFetchURLSources or null`
28502833 
2851 Which sources contribute to the set of URLs web fetch may fetch.
2834 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
28522835 
2853 Each key is a tagged variant: `user_input` is `all` or `none`; the
2854 two tool filters are `all`, `none`, `only` (only the named tools'
2855 results) or `except` (every result but the named tools'). A named tool
2856 must be declared in this request's `tools[]`.
2857 
28582836 - `use_cache: optional boolean`
28592837 
28602838 Whether to use cached content. Set to false to bypass the cache and fetch fresh content. Only set to false when the user explicitly requests fresh content or when fetching rapidly-changing sources.
from line 2877
28992877 
29002878 Maximum number of times the tool can be used in the API request.
29012879 
2902 exclusiveMinimum: 0
2880 minimum: 1
29032881 
29042882 - `response_inclusion: optional "full" or "excluded"`
29052883 
from line 2939
29612939 
29622940 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
29632941 
2964 exclusiveMinimum: 0
2942 minimum: 1
29652943 
29662944 - `max_uses: optional number or null`
29672945 
29682946 Maximum number of times the tool can be used in the API request.
29692947 
2970 exclusiveMinimum: 0
2948 minimum: 1
29712949 
29722950 - `response_inclusion: optional "full" or "excluded"`
29732951 
from line 2961
29832961 
29842962 - `url_sources: optional WebFetchURLSources or null`
29852963 
2986 Which sources contribute to the set of URLs web fetch may fetch.
2964 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
29872965 
2988 Each key is a tagged variant: `user_input` is `all` or `none`; the
2989 two tool filters are `all`, `none`, `only` (only the named tools'
2990 results) or `except` (every result but the named tools'). A named tool
2991 must be declared in this request's `tools[]`.
2992 
29932966 - `use_cache: optional boolean`
29942967 
29952968 Whether to use cached content. Set to false to bypass the cache and fetch fresh content. Only set to false when the user explicitly requests fresh content or when fetching rapidly-changing sources.
from line 3049
30763049 
30773050 Note that even with `temperature` of `0.0`, the results will not be fully deterministic.
30783051 
3079 maximum: 1, minimum: 0
3052 minimum: 0, maximum: 1
30803053 
30813054- `top_k: optional number`
30823055 
from line 3073
31003073 
31013074 Recommended for advanced use cases only.
31023075 
3103 maximum: 1, minimum: 0
3076 minimum: 0, maximum: 1
31043077 
31053078### Returns
31063079 
from line 3095
31223095 
31233096 - `container: Container or null`
31243097 
3125 Information about the container used in the request (for the code execution tool)
3098 Information about the container used in this request.
31263099 
3100 This will be non-null if a container tool (e.g. code execution) was used.
3101 
31273102 - `id: string`
31283103 
31293104 Identifier for the container used in this request
from line 3125
31503125 
31513126 Skill ID
31523127 
3153 maxLength: 64, minLength: 1
3128 minLength: 1, maxLength: 64
31543129 
31553130 - `version: string`
31563131 
31573132 The resolved version: a skill version ID for custom skills.
31583133 
3159 maxLength: 64, minLength: 1
3134 minLength: 1, maxLength: 64
31603135 
31613136 - `content: array of ContentBlock`
31623137 
from line 3304
33293304 
33303305 - `text: string`
33313306 
3332 minLength: 0
3333 
33343307 - `ThinkingBlock object`
33353308 
33363309 - `type: "thinking"`
from line 3384
34113384 
34123385 For a toolset member tool_use, the toolset family.
34133386 
3414 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
3387 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
34153388 
34163389 - `ServerToolUseBlock object`
34173390 
from line 3832
38593832 
38603833 - `tool_name: string`
38613834 
3862 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
3835 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
38633836 
38643837 - `tool_use_id: string`
38653838 
from line 3850
38773850 
38783851 - `diagnostics: Diagnostics or null`
38793852 
3880 Request-level diagnostics: why the prompt cache could not fully reuse
3881 the prefix of the request named by `diagnostics.previous_message_id`.
3853 Request-level diagnostics. `null` when the request did not supply `diagnostics`, or when it did and no prompt-cache divergence was detected.
38823854 
38833855 - `cache_miss_reason: CacheMissReason or null`
38843856 
from line 3914
39423914 
39433915 See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.
39443916 
3917 - `string`
3918 
39453919 - `"claude-fable-5-1" or "claude-opus-5-5" or "claude-mythos-5-1" or 15 more`
39463920 
39473921 The model that will complete your prompt.
from line 3958
39843958 
39853959 Powerful intelligence for long-running agents and coding
39863960 
3987 - `"claude-mythos-preview"`
3988 
3989 New class of intelligence, strongest in coding and cybersecurity
3990 
39913961 - `"claude-opus-4-6"`
39923962 
39933963 Powerful intelligence for long-running agents and coding
from line 3990
40203990 
40213991 High-performance model for agents and coding
40223992 
4023 - `string`
3993 - `"claude-mythos-preview"`
40243994 
3995 **Deprecated**: Will reach end-of-life on June 30, 2026. Please migrate to claude-mythos-5. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.
3996 
3997 New class of intelligence, strongest in coding and cybersecurity
3998 
40253999 - `role: "assistant"`
40264000 
40274001 Conversational role of the generated message.
from line 4006
40324006 
40334007 - `stop_details: RefusalStopDetails or null`
40344008 
4035 Structured information about a refusal.
4009 Structured information about why model output stopped.
40364010 
4011 This is `null` when the `stop_reason` has no additional detail to report.
4012 
40374013 - `type: "refusal"`
40384014 
40394015 default: refusal
from line 4016
40404016 
40414017 - `category: "cyber" or "bio" or "frontier_llm" or 2 more or null`
40424018 
4043 The policy category that triggered a refusal.
4019 The policy category that triggered the refusal.
40444020 
4021 `null` when the refusal doesn't map to a named category.
4022 
40454023 - `"cyber"`
40464024 
40474025 The request could enable cyber harm, such as malware or exploit development. Benign cybersecurity work can also trigger this category.
from line 4206
42284206 
42294207 - `container: Container or null`
42304208 
4231 Information about the container used in the request (for the code execution tool)
4209 Information about the container used in this request.
42324210 
4211 This will be non-null if a container tool (e.g. code execution) was used.
4212 
42334213 - `stop_details: RefusalStopDetails or null`
42344214 
4235 Structured information about a refusal.
4215 Structured information about why model output stopped.
42364216 
4217 This is `null` when the `stop_reason` has no additional detail to report.
4218 
42374219 - `stop_reason: StopReason or null`
42384220 
42394221 - `stop_sequence: string or null`
from line 4624
46424624 
46434625 - `document_title: string or null`
46444626 
4645 maxLength: 500, minLength: 1
4627 minLength: 1, maxLength: 500
46464628 
46474629 - `end_char_index: number`
46484630 
from line 4644
46624644 
46634645 - `document_title: string or null`
46644646 
4665 maxLength: 500, minLength: 1
4647 minLength: 1, maxLength: 500
46664648 
46674649 - `end_page_number: number`
46684650 
from line 4668
46864668 
46874669 - `document_title: string or null`
46884670 
4689 maxLength: 500, minLength: 1
4671 minLength: 1, maxLength: 500
46904672 
46914673 - `end_block_index: number`
46924674 
from line 4692
47104692 
47114693 - `title: string or null`
47124694 
4713 maxLength: 512, minLength: 1
4695 minLength: 1, maxLength: 512
47144696 
47154697 - `url: string`
47164698 
from line 4848
48664848 
48674849 - `title: optional string or null`
48684850 
4869 maxLength: 500, minLength: 1
4851 minLength: 1, maxLength: 500
48704852 
48714853 - `SearchResultBlockParam object`
48724854 
from line 4912
49304912 
49314913 - `name: string`
49324914 
4933 maxLength: 200, minLength: 1
4915 minLength: 1, maxLength: 200
49344916 
49354917 - `cache_control: optional CacheControlEphemeral or null`
49364918 
from line 4948
49664948 
49674949 For a toolset member tool_use, the toolset family this member belongs to.
49684950 
4969 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
4951 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
49704952 
49714953 - `ToolResultBlockParam object`
49724954 
from line 4984
50024984 
50034985 - `tool_name: string`
50044986 
5005 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
4987 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
50064988 
50074989 - `cache_control: optional CacheControlEphemeral or null`
50084990 
from line 5012
50305012 
50315013 The caller-assigned identifier for this tab, unique within the inventory.
50325014 
5033 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
5015 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
50345016 
50355017 - `title: string`
50365018 
from line 5038
50565038 
50575039 Tabs opened and download state changes during this call. "Nothing to report" is expressed by omitting the field, never by an empty list.
50585040 
5059 maxItems: 200, minItems: 1
5041 minItems: 1, maxItems: 200
50605042 
50615043 - `BrowserStateChangeTabOpened object`
50625044 
from line 5056
50745056 
50755057 The `tab_id` of the opened tab, present in `tabs`.
50765058 
5077 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
5059 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
50785060 
50795061 - `BrowserStateChangeDownloadStarted object`
50805062 
from line 5068
50865068 
50875069 The caller-assigned identifier for this download, stable across the state changes reporting it.
50885070 
5089 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
5071 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
50905072 
50915073 - `url: string`
50925074 
from line 5089
51075089 
51085090 The caller-assigned identifier for this download, stable across the state changes reporting it.
51095091 
5110 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
5092 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
51115093 
51125094 - `url: string`
51135095 
from line 5101
51195101 
51205102 Where the executor saved the file, on the executor's filesystem. Only included when another tool in the same environment can read the file at that path.
51215103 
5122 pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$, maxLength: 4096
5104 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
51235105 
51245106 - `size_bytes: optional number or null`
51255107 
from line 5119
51375119 
51385120 The caller-assigned identifier for this download, stable across the state changes reporting it.
51395121 
5140 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
5122 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
51415123 
51425124 - `url: string`
51435125 
from line 5131
51495131 
51505132 The failure or cancellation detail, when known.
51515133 
5152 pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$, maxLength: 4096
5134 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
51535135 
51545136 - `is_error: optional boolean`
51555137 
from line 5139
51575139 
51585140 For a toolset member tool_result, the toolset family of the paired tool_use.
51595141 
5160 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
5142 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
51615143 
51625144 - `ServerToolUseBlockParam object`
51635145 
from line 5517
55355517 
55365518 - `tool_name: string`
55375519 
5538 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
5520 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
55395521 
55405522 - `cache_control: optional CacheControlEphemeral or null`
55415523 
from line 5558
55765558 
55775559 See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.
55785560 
5561 - `string`
5562 
55795563 - `"claude-fable-5-1" or "claude-opus-5-5" or "claude-mythos-5-1" or 15 more`
55805564 
55815565 The model that will complete your prompt.
from line 5602
56185602 
56195603 Powerful intelligence for long-running agents and coding
56205604 
5621 - `"claude-mythos-preview"`
5622 
5623 New class of intelligence, strongest in coding and cybersecurity
5624 
56255605 - `"claude-opus-4-6"`
56265606 
56275607 Powerful intelligence for long-running agents and coding
from line 5634
56545634 
56555635 High-performance model for agents and coding
56565636 
5657 - `string`
5637 - `"claude-mythos-preview"`
56585638 
5639 **Deprecated**: Will reach end-of-life on June 30, 2026. Please migrate to claude-mythos-5. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.
5640 
5641 New class of intelligence, strongest in coding and cybersecurity
5642 
56595643- `cache_control: optional CacheControlEphemeral or null`
56605644 
56615645 Top-level cache control automatically applies a cache_control marker to the last cacheable block in the request.
from line 5650
56665650 
56675651 - `effort: optional "low" or "medium" or "high" or 2 more or null`
56685652 
5669 All possible effort levels.
5653 How much effort the model should put into its response. Higher effort levels may result in more thorough analysis but take longer.
56705654 
5655 Valid values are `low`, `medium`, `high`, `xhigh`, or `max`.
5656 
56715657 - `"low"`
56725658 
56735659 - `"medium"`
from line 5878
58925878 
58935879 This is how the tool will be called by the model and in `tool_use` blocks.
58945880 
5895 maxLength: 128, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,128}$
5881 minLength: 1, maxLength: 128, pattern: ^[a-zA-Z0-9_-]{1,128}$
58965882 
58975883 - `allowed_callers: optional array of "direct" or "code_execution_20250825" or "code_execution_20260120" or "code_execution_20260521"`
58985884 
from line 6095
61096095 
61106096 - `configs: optional BrowserToolsetConfigs or null`
61116097 
6112 Per-member configuration for `browser_toolset_20260801`: one
6113 optional field per member tool, keyed by the member name — the same
6114 name the member's `tool_use` blocks carry. Every member is an
6115 accepted key, and a member's defaults apply wherever its key is
6116 absent. Unknown keys are rejected: the field set is this toolset
6117 version's complete member set.
6098 Sparse per-member overrides, keyed by member name. Absent, null, and {} are equivalent; a member's defaults apply wherever its key is absent.
61186099 
61196100 - `type: optional BrowserTypeConfig or null`
61206101 
from line 6522
65416522 
65426523 - `configs: optional ComputerToolsetConfigs or null`
65436524 
6544 Per-member configuration for `computer_toolset_20260801`: one
6545 optional field per member tool, keyed by the member name — the same
6546 name the member's `tool_use` blocks carry. Every member is an
6547 accepted key, and a member's defaults apply wherever its key is
6548 absent. Unknown keys are rejected: the field set is this toolset
6549 version's complete member set.
6525 Sparse per-member overrides, keyed by member name. Absent, null, and {} are equivalent; a member's defaults apply wherever its key is absent.
65506526 
65516527 - `type: optional ComputerTypeConfig or null`
65526528 
from line 6876
69006876 
69016877 Maximum number of times the tool can be used in the API request.
69026878 
6903 exclusiveMinimum: 0
6879 minimum: 1
69046880 
69056881 - `strict: optional boolean`
69066882 
from line 6892
69166892 
69176893 The city of the user.
69186894 
6919 maxLength: 255, minLength: 1
6895 minLength: 1, maxLength: 255
69206896 
69216897 - `country: optional string or null`
69226898 
69236899 The two letter [ISO country code](https://en.wikipedia.org/wiki/ISO_3166-1_alpha-2) of the user.
69246900 
6925 maxLength: 2, minLength: 2
6901 minLength: 2, maxLength: 2
69266902 
69276903 - `region: optional string or null`
69286904 
69296905 The region of the user.
69306906 
6931 maxLength: 255, minLength: 1
6907 minLength: 1, maxLength: 255
69326908 
69336909 - `timezone: optional string or null`
69346910 
69356911 The [IANA timezone](https://nodatime.org/TimeZones) of the user.
69366912 
6937 maxLength: 255, minLength: 1
6913 minLength: 1, maxLength: 255
69386914 
69396915 - `WebFetchTool20250910 object`
69406916 
from line 6956
69806956 
69816957 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
69826958 
6983 exclusiveMinimum: 0
6959 minimum: 1
69846960 
69856961 - `max_uses: optional number or null`
69866962 
69876963 Maximum number of times the tool can be used in the API request.
69886964 
6989 exclusiveMinimum: 0
6965 minimum: 1
69906966 
69916967 - `strict: optional boolean`
69926968 
from line 6970
69946970 
69956971 - `url_sources: optional WebFetchURLSources or null`
69966972 
6997 Which sources contribute to the set of URLs web fetch may fetch.
6973 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
69986974 
6999 Each key is a tagged variant: `user_input` is `all` or `none`; the
7000 two tool filters are `all`, `none`, `only` (only the named tools'
7001 results) or `except` (every result but the named tools'). A named tool
7002 must be declared in this request's `tools[]`.
7003 
70046975 - `client_tool_results: optional WebFetchURLSourceAll or WebFetchURLSourceNone or WebFetchURLSourceOnly or WebFetchURLSourceExcept`
70056976 
70066977 Which client tools' results contribute fetchable URLs: "all", "none", or an only or except list of client tool names from tools[].
from line 7094
71237094 
71247095 Maximum number of times the tool can be used in the API request.
71257096 
7126 exclusiveMinimum: 0
7097 minimum: 1
71277098 
71287099 - `strict: optional boolean`
71297100 
from line 7148
71777148 
71787149 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
71797150 
7180 exclusiveMinimum: 0
7151 minimum: 1
71817152 
71827153 - `max_uses: optional number or null`
71837154 
71847155 Maximum number of times the tool can be used in the API request.
71857156 
7186 exclusiveMinimum: 0
7157 minimum: 1
71877158 
71887159 - `strict: optional boolean`
71897160 
from line 7162
71917162 
71927163 - `url_sources: optional WebFetchURLSources or null`
71937164 
7194 Which sources contribute to the set of URLs web fetch may fetch.
7165 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
71957166 
7196 Each key is a tagged variant: `user_input` is `all` or `none`; the
7197 two tool filters are `all`, `none`, `only` (only the named tools'
7198 results) or `except` (every result but the named tools'). A named tool
7199 must be declared in this request's `tools[]`.
7200 
72017167 - `WebFetchTool20260309 object`
72027168 
72037169 Web fetch tool with use_cache parameter for bypassing cached content.
from line 7210
72447210 
72457211 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
72467212 
7247 exclusiveMinimum: 0
7213 minimum: 1
72487214 
72497215 - `max_uses: optional number or null`
72507216 
72517217 Maximum number of times the tool can be used in the API request.
72527218 
7253 exclusiveMinimum: 0
7219 minimum: 1
72547220 
72557221 - `strict: optional boolean`
72567222 
from line 7224
72587224 
72597225 - `url_sources: optional WebFetchURLSources or null`
72607226 
7261 Which sources contribute to the set of URLs web fetch may fetch.
7227 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
72627228 
7263 Each key is a tagged variant: `user_input` is `all` or `none`; the
7264 two tool filters are `all`, `none`, `only` (only the named tools'
7265 results) or `except` (every result but the named tools'). A named tool
7266 must be declared in this request's `tools[]`.
7267 
72687229 - `use_cache: optional boolean`
72697230 
72707231 Whether to use cached content. Set to false to bypass the cache and fetch fresh content. Only set to false when the user explicitly requests fresh content or when fetching rapidly-changing sources.
from line 7270
73097270 
73107271 Maximum number of times the tool can be used in the API request.
73117272 
7312 exclusiveMinimum: 0
7273 minimum: 1
73137274 
73147275 - `response_inclusion: optional "full" or "excluded"`
73157276 
from line 7332
73717332 
73727333 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
73737334 
7374 exclusiveMinimum: 0
7335 minimum: 1
73757336 
73767337 - `max_uses: optional number or null`
73777338 
73787339 Maximum number of times the tool can be used in the API request.
73797340 
7380 exclusiveMinimum: 0
7341 minimum: 1
73817342 
73827343 - `response_inclusion: optional "full" or "excluded"`
73837344 
from line 7354
73937354 
73947355 - `url_sources: optional WebFetchURLSources or null`
73957356 
7396 Which sources contribute to the set of URLs web fetch may fetch.
7357 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
73977358 
7398 Each key is a tagged variant: `user_input` is `all` or `none`; the
7399 two tool filters are `all`, `none`, `only` (only the named tools'
7400 results) or `except` (every result but the named tools'). A named tool
7401 must be declared in this request's `tools[]`.
7402 
74037359 - `use_cache: optional boolean`
74047360 
74057361 Whether to use cached content. Set to false to bypass the cache and fetch fresh content. Only set to false when the user explicitly requests fresh content or when fetching rapidly-changing sources.
from line 8139
81838139 
81848140 The caller-assigned identifier for this tab, unique within the inventory.
81858141 
8186 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
8142 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
81878143 
81888144 - `title: string`
81898145 
from line 8182
82268182 
82278183 Tabs opened and download state changes during this call. "Nothing to report" is expressed by omitting the field, never by an empty list.
82288184 
8229 maxItems: 200, minItems: 1
8185 minItems: 1, maxItems: 200
82308186 
82318187 - `BrowserStateChangeTabOpened object`
82328188 
from line 8200
82448200 
82458201 The `tab_id` of the opened tab, present in `tabs`.
82468202 
8247 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
8203 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
82488204 
82498205 - `BrowserStateChangeDownloadStarted object`
82508206 
from line 8212
82568212 
82578213 The caller-assigned identifier for this download, stable across the state changes reporting it.
82588214 
8259 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
8215 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
82608216 
82618217 - `url: string`
82628218 
from line 8233
82778233 
82788234 The caller-assigned identifier for this download, stable across the state changes reporting it.
82798235 
8280 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
8236 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
82818237 
82828238 - `url: string`
82838239 
from line 8245
82898245 
82908246 Where the executor saved the file, on the executor's filesystem. Only included when another tool in the same environment can read the file at that path.
82918247 
8292 pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$, maxLength: 4096
8248 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
82938249 
82948250 - `size_bytes: optional number or null`
82958251 
from line 8263
83078263 
83088264 The caller-assigned identifier for this download, stable across the state changes reporting it.
83098265 
8310 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
8266 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
83118267 
83128268 - `url: string`
83138269 
from line 8275
83198275 
83208276 The failure or cancellation detail, when known.
83218277 
8322 pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$, maxLength: 4096
8278 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
83238279 
83248280### Browser State Change
83258281 
from line 8297
83418297 
83428298 The `tab_id` of the opened tab, present in `tabs`.
83438299 
8344 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
8300 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
83458301 
83468302 - `BrowserStateChangeDownloadStarted object`
83478303 
from line 8309
83538309 
83548310 The caller-assigned identifier for this download, stable across the state changes reporting it.
83558311 
8356 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
8312 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
83578313 
83588314 - `url: string`
83598315 
from line 8330
83748330 
83758331 The caller-assigned identifier for this download, stable across the state changes reporting it.
83768332 
8377 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
8333 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
83788334 
83798335 - `url: string`
83808336 
from line 8342
83868342 
83878343 Where the executor saved the file, on the executor's filesystem. Only included when another tool in the same environment can read the file at that path.
83888344 
8389 pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$, maxLength: 4096
8345 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
83908346 
83918347 - `size_bytes: optional number or null`
83928348 
from line 8360
84048360 
84058361 The caller-assigned identifier for this download, stable across the state changes reporting it.
84068362 
8407 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
8363 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
84088364 
8409
8365 - `url: string`
8366 
8367 The final post-redirect URL the download was served from.
8368 
8369 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
8370 
8371 - `error: optional string or null`
8372 
8373 The failure or cancellation detail, when known.
8374 
8375 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
8376 
8377### Browser State Change Download Completed
8378 
8379- `BrowserStateChangeDownloadCompleted object`
8380 
8381 A file download that finished during this call, reported with the
8382 same `download_id` as its `download_started` — or without a prior
8383 `download_started`, when the download finished during the call that
8384 started it (at most one state change per `download_id` per result).
8385 
8386 - `type: "download_completed"`
8387 
8388 - `download_id: string`
8389 
8390 The caller-assigned identifier for this download, stable across the state changes reporting it.
8391 
8392 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
8393 
8394 - `url: string`
8395 
8396 The final post-redirect URL the download was served from.
8397 
8398 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
8399 
8400 - `path: optional string or null`
8401 
8402 Where the executor saved the file, on the executor's filesystem. Only included when another tool in the same environment can read the file at that path.
8403 
8404 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
8405 
8406 - `size_bytes: optional number or null`
8407 
8408 The completed download's size.
8409 
8410 minimum: 0
8411 
8412### Browser State Change Download Failed
8413 
8414- `BrowserStateChangeDownloadFailed object`
8415 
8416 A file download that failed — or was cancelled — during this call.
8417 
8418 - `type: "download_failed"`
8419 
8420 - `download_id: string`
8421 
8422 The caller-assigned identifier for this download, stable across the state changes reporting it.
8423 
8424 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x

api/messages/batches Changed · +326 / -333 lines

This page is larger than the 256 KiB this site keeps, so one side of the diff below stops where the stored text does.

from line 33
3333 
3434 List of requests for prompt completion. Each is an individual request to create a Message.
3535 
36 maxItems: 100000, minItems: 1
36 minItems: 1, maxItems: 100000
3737 
3838 - `custom_id: string`
3939 
from line 41
4141 
4242 Must be unique for each request within the Message Batch.
4343 
44 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,64}$
44 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]{1,64}$
4545 
4646 - `params: object`
4747 
from line 161
161161 
162162 - `document_title: string or null`
163163 
164 maxLength: 500, minLength: 1
164 minLength: 1, maxLength: 500
165165 
166166 - `end_char_index: number`
167167 
from line 181
181181 
182182 - `document_title: string or null`
183183 
184 maxLength: 500, minLength: 1
184 minLength: 1, maxLength: 500
185185 
186186 - `end_page_number: number`
187187 
from line 205
205205 
206206 - `document_title: string or null`
207207 
208 maxLength: 500, minLength: 1
208 minLength: 1, maxLength: 500
209209 
210210 - `end_block_index: number`
211211 
from line 229
229229 
230230 - `title: string or null`
231231 
232 maxLength: 512, minLength: 1
232 minLength: 1, maxLength: 512
233233 
234234 - `url: string`
235235 
from line 385
385385 
386386 - `title: optional string or null`
387387 
388 maxLength: 500, minLength: 1
388 minLength: 1, maxLength: 500
389389 
390390 - `SearchResultBlockParam object`
391391 
from line 449
449449 
450450 - `name: string`
451451 
452 maxLength: 200, minLength: 1
452 minLength: 1, maxLength: 200
453453 
454454 - `cache_control: optional CacheControlEphemeral or null`
455455 
from line 485
485485 
486486 For a toolset member tool_use, the toolset family this member belongs to.
487487 
488 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
488 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
489489 
490490 - `ToolResultBlockParam object`
491491 
from line 521
521521 
522522 - `tool_name: string`
523523 
524 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
524 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
525525 
526526 - `cache_control: optional CacheControlEphemeral or null`
527527 
from line 549
549549 
550550 The caller-assigned identifier for this tab, unique within the inventory.
551551 
552 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
552 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
553553 
554554 - `title: string`
555555 
from line 575
575575 
576576 Tabs opened and download state changes during this call. "Nothing to report" is expressed by omitting the field, never by an empty list.
577577 
578 maxItems: 200, minItems: 1
578 minItems: 1, maxItems: 200
579579 
580580 - `BrowserStateChangeTabOpened object`
581581 
from line 593
593593 
594594 The `tab_id` of the opened tab, present in `tabs`.
595595 
596 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
596 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
597597 
598598 - `BrowserStateChangeDownloadStarted object`
599599 
from line 605
605605 
606606 The caller-assigned identifier for this download, stable across the state changes reporting it.
607607 
608 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
608 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
609609 
610610 - `url: string`
611611 
from line 626
626626 
627627 The caller-assigned identifier for this download, stable across the state changes reporting it.
628628 
629 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
629 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
630630 
631631 - `url: string`
632632 
from line 638
638638 
639639 Where the executor saved the file, on the executor's filesystem. Only included when another tool in the same environment can read the file at that path.
640640 
641 pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$, maxLength: 4096
641 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
642642 
643643 - `size_bytes: optional number or null`
644644 
from line 656
656656 
657657 The caller-assigned identifier for this download, stable across the state changes reporting it.
658658 
659 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
659 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
660660 
661661 - `url: string`
662662 
from line 668
668668 
669669 The failure or cancellation detail, when known.
670670 
671 pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$, maxLength: 4096
671 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
672672 
673673 - `is_error: optional boolean`
674674 
from line 676
676676 
677677 For a toolset member tool_result, the toolset family of the paired tool_use.
678678 
679 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
679 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
680680 
681681 - `ServerToolUseBlockParam object`
682682 
from line 1054
10541054 
10551055 - `tool_name: string`
10561056 
1057 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
1057 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
10581058 
10591059 - `cache_control: optional CacheControlEphemeral or null`
10601060 
from line 1095
10951095 
10961096 See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.
10971097 
1098 - `string`
1099 
10981100 - `"claude-fable-5-1" or "claude-opus-5-5" or "claude-mythos-5-1" or 15 more`
10991101 
11001102 The model that will complete your prompt.
from line 1139
11371139 
11381140 Powerful intelligence for long-running agents and coding
11391141 
1140 - `"claude-mythos-preview"`
1141 
1142 New class of intelligence, strongest in coding and cybersecurity
1143 
11441142 - `"claude-opus-4-6"`
11451143 
11461144 Powerful intelligence for long-running agents and coding
from line 1171
11731171 
11741172 High-performance model for agents and coding
11751173 
1176 - `string`
1174 - `"claude-mythos-preview"`
11771175 
1176 **Deprecated**: Will reach end-of-life on June 30, 2026. Please migrate to claude-mythos-5. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.
1177 
1178 New class of intelligence, strongest in coding and cybersecurity
1179 
11781180 - `cache_control: optional CacheControlEphemeral or null`
11791181 
11801182 Top-level cache control automatically applies a cache_control marker to the last cacheable block in the request.
from line 1211
12091211 
12101212 Skill ID
12111213 
1212 maxLength: 64, minLength: 1
1214 minLength: 1, maxLength: 64
12131215 
12141216 - `version: optional string`
12151217 
12161218 Skill version or 'latest' for most recent version
12171219 
1218 maxLength: 64, minLength: 1
1220 minLength: 1, maxLength: 64
12191221 
12201222 - `string`
12211223 
12221224 - `diagnostics: optional DiagnosticsParam or null`
12231225 
1224 Request-level diagnostics. Currently carries the previous response
1225 id for prompt-cache divergence reporting.
1226 Request-level diagnostics. Supply `previous_message_id` to have the response include `diagnostics.cache_miss_reason` explaining any prompt-cache divergence from that prior request.
12261227 
12271228 - `previous_message_id: optional string or null`
12281229 
from line 1253
12521253 
12531254 - `effort: optional "low" or "medium" or "high" or 2 more or null`
12541255 
1255 All possible effort levels.
1256 How much effort the model should put into its response. Higher effort levels may result in more thorough analysis but take longer.
12561257 
1258 Valid values are `low`, `medium`, `high`, `xhigh`, or `max`.
1259 
12571260 - `"low"`
12581261 
12591262 - `"medium"`
from line 1505
15021505 
15031506 This is how the tool will be called by the model and in `tool_use` blocks.
15041507 
1505 maxLength: 128, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,128}$
1508 minLength: 1, maxLength: 128, pattern: ^[a-zA-Z0-9_-]{1,128}$
15061509 
15071510 - `allowed_callers: optional array of "direct" or "code_execution_20250825" or "code_execution_20260120" or "code_execution_20260521"`
15081511 
from line 1722
17191722 
17201723 - `configs: optional BrowserToolsetConfigs or null`
17211724 
1722 Per-member configuration for `browser_toolset_20260801`: one
1723 optional field per member tool, keyed by the member name — the same
1724 name the member's `tool_use` blocks carry. Every member is an
1725 accepted key, and a member's defaults apply wherever its key is
1726 absent. Unknown keys are rejected: the field set is this toolset
1727 version's complete member set.
1725 Sparse per-member overrides, keyed by member name. Absent, null, and {} are equivalent; a member's defaults apply wherever its key is absent.
17281726 
17291727 - `type: optional BrowserTypeConfig or null`
17301728 
from line 2149
21512149 
21522150 - `configs: optional ComputerToolsetConfigs or null`
21532151 
2154 Per-member configuration for `computer_toolset_20260801`: one
2155 optional field per member tool, keyed by the member name — the same
2156 name the member's `tool_use` blocks carry. Every member is an
2157 accepted key, and a member's defaults apply wherever its key is
2158 absent. Unknown keys are rejected: the field set is this toolset
2159 version's complete member set.
2152 Sparse per-member overrides, keyed by member name. Absent, null, and {} are equivalent; a member's defaults apply wherever its key is absent.
21602153 
21612154 - `type: optional ComputerTypeConfig or null`
21622155 
from line 2503
25102503 
25112504 Maximum number of times the tool can be used in the API request.
25122505 
2513 exclusiveMinimum: 0
2506 minimum: 1
25142507 
25152508 - `strict: optional boolean`
25162509 
from line 2519
25262519 
25272520 The city of the user.
25282521 
2529 maxLength: 255, minLength: 1
2522 minLength: 1, maxLength: 255
25302523 
25312524 - `country: optional string or null`
25322525 
25332526 The two letter [ISO country code](https://en.wikipedia.org/wiki/ISO_3166-1_alpha-2) of the user.
25342527 
2535 maxLength: 2, minLength: 2
2528 minLength: 2, maxLength: 2
25362529 
25372530 - `region: optional string or null`
25382531 
25392532 The region of the user.
25402533 
2541 maxLength: 255, minLength: 1
2534 minLength: 1, maxLength: 255
25422535 
25432536 - `timezone: optional string or null`
25442537 
25452538 The [IANA timezone](https://nodatime.org/TimeZones) of the user.
25462539 
2547 maxLength: 255, minLength: 1
2540 minLength: 1, maxLength: 255
25482541 
25492542 - `WebFetchTool20250910 object`
25502543 
from line 2583
25902583 
25912584 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
25922585 
2593 exclusiveMinimum: 0
2586 minimum: 1
25942587 
25952588 - `max_uses: optional number or null`
25962589 
25972590 Maximum number of times the tool can be used in the API request.
25982591 
2599 exclusiveMinimum: 0
2592 minimum: 1
26002593 
26012594 - `strict: optional boolean`
26022595 
from line 2597
26042597 
26052598 - `url_sources: optional WebFetchURLSources or null`
26062599 
2607 Which sources contribute to the set of URLs web fetch may fetch.
2600 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
26082601 
2609 Each key is a tagged variant: `user_input` is `all` or `none`; the
2610 two tool filters are `all`, `none`, `only` (only the named tools'
2611 results) or `except` (every result but the named tools'). A named tool
2612 must be declared in this request's `tools[]`.
2613 
26142602 - `client_tool_results: optional WebFetchURLSourceAll or WebFetchURLSourceNone or WebFetchURLSourceOnly or WebFetchURLSourceExcept`
26152603 
26162604 Which client tools' results contribute fetchable URLs: "all", "none", or an only or except list of client tool names from tools[].
from line 2721
27332721 
27342722 Maximum number of times the tool can be used in the API request.
27352723 
2736 exclusiveMinimum: 0
2724 minimum: 1
27372725 
27382726 - `strict: optional boolean`
27392727 
from line 2775
27872775 
27882776 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
27892777 
2790 exclusiveMinimum: 0
2778 minimum: 1
27912779 
27922780 - `max_uses: optional number or null`
27932781 
27942782 Maximum number of times the tool can be used in the API request.
27952783 
2796 exclusiveMinimum: 0
2784 minimum: 1
27972785 
27982786 - `strict: optional boolean`
27992787 
from line 2789
28012789 
28022790 - `url_sources: optional WebFetchURLSources or null`
28032791 
2804 Which sources contribute to the set of URLs web fetch may fetch.
2792 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
28052793 
2806 Each key is a tagged variant: `user_input` is `all` or `none`; the
2807 two tool filters are `all`, `none`, `only` (only the named tools'
2808 results) or `except` (every result but the named tools'). A named tool
2809 must be declared in this request's `tools[]`.
2810 
28112794 - `WebFetchTool20260309 object`
28122795 
28132796 Web fetch tool with use_cache parameter for bypassing cached content.
from line 2837
28542837 
28552838 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
28562839 
2857 exclusiveMinimum: 0
2840 minimum: 1
28582841 
28592842 - `max_uses: optional number or null`
28602843 
28612844 Maximum number of times the tool can be used in the API request.
28622845 
2863 exclusiveMinimum: 0
2846 minimum: 1
28642847 
28652848 - `strict: optional boolean`
28662849 
from line 2851
28682851 
28692852 - `url_sources: optional WebFetchURLSources or null`
28702853 
2871 Which sources contribute to the set of URLs web fetch may fetch.
2854 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
28722855 
2873 Each key is a tagged variant: `user_input` is `all` or `none`; the
2874 two tool filters are `all`, `none`, `only` (only the named tools'
2875 results) or `except` (every result but the named tools'). A named tool
2876 must be declared in this request's `tools[]`.
2877 
28782856 - `use_cache: optional boolean`
28792857 
28802858 Whether to use cached content. Set to false to bypass the cache and fetch fresh content. Only set to false when the user explicitly requests fresh content or when fetching rapidly-changing sources.
from line 2897
29192897 
29202898 Maximum number of times the tool can be used in the API request.
29212899 
2922 exclusiveMinimum: 0
2900 minimum: 1
29232901 
29242902 - `response_inclusion: optional "full" or "excluded"`
29252903 
from line 2959
29812959 
29822960 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
29832961 
2984 exclusiveMinimum: 0
2962 minimum: 1
29852963 
29862964 - `max_uses: optional number or null`
29872965 
29882966 Maximum number of times the tool can be used in the API request.
29892967 
2990 exclusiveMinimum: 0
2968 minimum: 1
29912969 
29922970 - `response_inclusion: optional "full" or "excluded"`
29932971 
from line 2981
30032981 
30042982 - `url_sources: optional WebFetchURLSources or null`
30052983 
3006 Which sources contribute to the set of URLs web fetch may fetch.
2984 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
30072985 
3008 Each key is a tagged variant: `user_input` is `all` or `none`; the
3009 two tool filters are `all`, `none`, `only` (only the named tools'
3010 results) or `except` (every result but the named tools'). A named tool
3011 must be declared in this request's `tools[]`.
3012 
30132986 - `use_cache: optional boolean`
30142987 
30152988 Whether to use cached content. Set to false to bypass the cache and fetch fresh content. Only set to false when the user explicitly requests fresh content or when fetching rapidly-changing sources.
from line 3069
30963069 
30973070 Note that even with `temperature` of `0.0`, the results will not be fully deterministic.
30983071 
3099 maximum: 1, minimum: 0
3072 minimum: 0, maximum: 1
31003073 
31013074 - `top_k: optional number`
31023075 
from line 3093
31203093 
31213094 Recommended for advanced use cases only.
31223095 
3123 maximum: 1, minimum: 0
3096 minimum: 0, maximum: 1
31243097 
31253098### Returns
31263099 
from line 3441
34683441 
34693442 Defaults to `20`. Ranges from `1` to `1000`.
34703443 
3471 default: 20, maximum: 1000, minimum: 1
3444 default: 20, minimum: 1, maximum: 1000
34723445 
34733446### Headers
34743447 
from line 3901
39283901 
39293902 - `container: Container or null`
39303903 
3931 Information about the container used in the request (for the code execution tool)
3904 Information about the container used in this request.
39323905 
3906 This will be non-null if a container tool (e.g. code execution) was used.
3907 
39333908 - `id: string`
39343909 
39353910 Identifier for the container used in this request
from line 3931
39563931 
39573932 Skill ID
39583933 
3959 maxLength: 64, minLength: 1
3934 minLength: 1, maxLength: 64
39603935 
39613936 - `version: string`
39623937 
39633938 The resolved version: a skill version ID for custom skills.
39643939 
3965 maxLength: 64, minLength: 1
3940 minLength: 1, maxLength: 64
39663941 
39673942 - `content: array of ContentBlock`
39683943 
from line 4110
41354110 
41364111 - `text: string`
41374112 
4138 minLength: 0
4139 
41404113 - `ThinkingBlock object`
41414114 
41424115 - `type: "thinking"`
from line 4190
42174190 
42184191 For a toolset member tool_use, the toolset family.
42194192 
4220 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
4193 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
42214194 
42224195 - `ServerToolUseBlock object`
42234196 
from line 4638
46654638 
46664639 - `tool_name: string`
46674640 
4668 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
4641 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
46694642 
46704643 - `tool_use_id: string`
46714644 
from line 4656
46834656 
46844657 - `diagnostics: Diagnostics or null`
46854658 
4686 Request-level diagnostics: why the prompt cache could not fully reuse
4687 the prefix of the request named by `diagnostics.previous_message_id`.
4659 Request-level diagnostics. `null` when the request did not supply `diagnostics`, or when it did and no prompt-cache divergence was detected.
46884660 
46894661 - `cache_miss_reason: CacheMissReason or null`
46904662 
from line 4720
47484720 
47494721 See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.
47504722 
4723 - `string`
4724 
47514725 - `"claude-fable-5-1" or "claude-opus-5-5" or "claude-mythos-5-1" or 15 more`
47524726 
47534727 The model that will complete your prompt.
from line 4764
47904764 
47914765 Powerful intelligence for long-running agents and coding
47924766 
4793 - `"claude-mythos-preview"`
4794 
4795 New class of intelligence, strongest in coding and cybersecurity
4796 
47974767 - `"claude-opus-4-6"`
47984768 
47994769 Powerful intelligence for long-running agents and coding
from line 4796
48264796 
48274797 High-performance model for agents and coding
48284798 
4829 - `string`
4799 - `"claude-mythos-preview"`
48304800 
4801 **Deprecated**: Will reach end-of-life on June 30, 2026. Please migrate to claude-mythos-5. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.
4802 
4803 New class of intelligence, strongest in coding and cybersecurity
4804 
48314805 - `role: "assistant"`
48324806 
48334807 Conversational role of the generated message.
from line 4812
48384812 
48394813 - `stop_details: RefusalStopDetails or null`
48404814 
4841 Structured information about a refusal.
4815 Structured information about why model output stopped.
48424816 
4817 This is `null` when the `stop_reason` has no additional detail to report.
4818 
48434819 - `type: "refusal"`
48444820 
48454821 default: refusal
from line 4822
48464822 
48474823 - `category: "cyber" or "bio" or "frontier_llm" or 2 more or null`
48484824 
4849 The policy category that triggered a refusal.
4825 The policy category that triggered the refusal.
48504826 
4827 `null` when the refusal doesn't map to a named category.
4828 
48514829 - `"cyber"`
48524830 
48534831 The request could enable cyber harm, such as malware or exploit development. Benign cybersecurity work can also trigger this category.
from line 5412
54345412 
54355413 - `container: Container or null`
54365414 
5437 Information about the container used in the request (for the code execution tool)
5415 Information about the container used in this request.
54385416 
5417 This will be non-null if a container tool (e.g. code execution) was used.
5418 
54395419 - `id: string`
54405420 
54415421 Identifier for the container used in this request
from line 5442
54625442 
54635443 Skill ID
54645444 
5465 maxLength: 64, minLength: 1
5445 minLength: 1, maxLength: 64
54665446 
54675447 - `version: string`
54685448 
54695449 The resolved version: a skill version ID for custom skills.
54705450 
5471 maxLength: 64, minLength: 1
5451 minLength: 1, maxLength: 64
54725452 
54735453 - `content: array of ContentBlock`
54745454 
from line 5621
56415621 
56425622 - `text: string`
56435623 
5644 minLength: 0
5645 
56465624 - `ThinkingBlock object`
56475625 
56485626 - `type: "thinking"`
from line 5701
57235701 
57245702 For a toolset member tool_use, the toolset family.
57255703 
5726 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
5704 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
57275705 
57285706 - `ServerToolUseBlock object`
57295707 
from line 6149
61716149 
61726150 - `tool_name: string`
61736151 
6174 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
6152 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
61756153 
61766154 - `tool_use_id: string`
61776155 
from line 6167
61896167 
61906168 - `diagnostics: Diagnostics or null`
61916169 
6192 Request-level diagnostics: why the prompt cache could not fully reuse
6193 the prefix of the request named by `diagnostics.previous_message_id`.
6170 Request-level diagnostics. `null` when the request did not supply `diagnostics`, or when it did and no prompt-cache divergence was detected.
61946171 
61956172 - `cache_miss_reason: CacheMissReason or null`
61966173 
from line 6231
62546231 
62556232 See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.
62566233 
6234 - `string`
6235 
62576236 - `"claude-fable-5-1" or "claude-opus-5-5" or "claude-mythos-5-1" or 15 more`
62586237 
62596238 The model that will complete your prompt.
from line 6275
62966275 
62976276 Powerful intelligence for long-running agents and coding
62986277 
6299 - `"claude-mythos-preview"`
6300 
6301 New class of intelligence, strongest in coding and cybersecurity
6302 
63036278 - `"claude-opus-4-6"`
63046279 
63056280 Powerful intelligence for long-running agents and coding
from line 6307
63326307 
63336308 High-performance model for agents and coding
63346309 
6335 - `string`
6310 - `"claude-mythos-preview"`
63366311 
6312 **Deprecated**: Will reach end-of-life on June 30, 2026. Please migrate to claude-mythos-5. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.
6313 
6314 New class of intelligence, strongest in coding and cybersecurity
6315 
63376316 - `role: "assistant"`
63386317 
63396318 Conversational role of the generated message.
from line 6323
63446323 
63456324 - `stop_details: RefusalStopDetails or null`
63466325 
6347 Structured information about a refusal.
6326 Structured information about why model output stopped.
63486327 
6328 This is `null` when the `stop_reason` has no additional detail to report.
6329 
63496330 - `type: "refusal"`
63506331 
63516332 default: refusal
from line 6333
63526333 
63536334 - `category: "cyber" or "bio" or "frontier_llm" or 2 more or null`
63546335 
6355 The policy category that triggered a refusal.
6336 The policy category that triggered the refusal.
63566337 
6338 `null` when the refusal doesn't map to a named category.
6339 
63576340 - `"cyber"`
63586341 
63596342 The request could enable cyber harm, such as malware or exploit development. Benign cybersecurity work can also trigger this category.
from line 6695
67126695 
67136696 - `container: Container or null`
67146697 
6715 Information about the container used in the request (for the code execution tool)
6698 Information about the container used in this request.
67166699 
6700 This will be non-null if a container tool (e.g. code execution) was used.
6701 
67176702 - `id: string`
67186703 
67196704 Identifier for the container used in this request
from line 6725
67406725 
67416726 Skill ID
67426727 
6743 maxLength: 64, minLength: 1
6728 minLength: 1, maxLength: 64
67446729 
67456730 - `version: string`
67466731 
67476732 The resolved version: a skill version ID for custom skills.
67486733 
6749 maxLength: 64, minLength: 1
6734 minLength: 1, maxLength: 64
67506735 
67516736 - `content: array of ContentBlock`
67526737 
from line 6904
69196904 
69206905 - `text: string`
69216906 
6922 minLength: 0
6923 
69246907 - `ThinkingBlock object`
69256908 
69266909 - `type: "thinking"`
from line 6984
70016984 
70026985 For a toolset member tool_use, the toolset family.
70036986 
7004 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
6987 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
70056988 
70066989 - `ServerToolUseBlock object`
70076990 
from line 7432
74497432 
74507433 - `tool_name: string`
74517434 
7452 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
7435 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
74537436 
74547437 - `tool_use_id: string`
74557438 
from line 7450
74677450 
74687451 - `diagnostics: Diagnostics or null`
74697452 
7470 Request-level diagnostics: why the prompt cache could not fully reuse
7471 the prefix of the request named by `diagnostics.previous_message_id`.
7453 Request-level diagnostics. `null` when the request did not supply `diagnostics`, or when it did and no prompt-cache divergence was detected.
74727454 
74737455 - `cache_miss_reason: CacheMissReason or null`
74747456 
from line 7514
75327514 
75337515 See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.
75347516 
7517 - `string`
7518 
75357519 - `"claude-fable-5-1" or "claude-opus-5-5" or "claude-mythos-5-1" or 15 more`
75367520 
75377521 The model that will complete your prompt.
from line 7558
75747558 
75757559 Powerful intelligence for long-running agents and coding
75767560 
7577 - `"claude-mythos-preview"`
7578 
7579 New class of intelligence, strongest in coding and cybersecurity
7580 
75817561 - `"claude-opus-4-6"`
75827562 
75837563 Powerful intelligence for long-running agents and coding
from line 7590
76107590 
76117591 High-performance model for agents and coding
76127592 
7613 - `string`
7593 - `"claude-mythos-preview"`
76147594 
7595 **Deprecated**: Will reach end-of-life on June 30, 2026. Please migrate to claude-mythos-5. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.
7596 
7597 New class of intelligence, strongest in coding and cybersecurity
7598 
76157599 - `role: "assistant"`
76167600 
76177601 Conversational role of the generated message.
from line 7606
76227606 
76237607 - `stop_details: RefusalStopDetails or null`
76247608 
7625 Structured information about a refusal.
7609 Structured information about why model output stopped.
76267610 
7611 This is `null` when the `stop_reason` has no additional detail to report.
7612 
76277613 - `type: "refusal"`
76287614 
76297615 default: refusal
from line 7616
76307616 
76317617 - `category: "cyber" or "bio" or "frontier_llm" or 2 more or null`
76327618 
7633 The policy category that triggered a refusal.
7619 The policy category that triggered the refusal.
76347620 
7621 `null` when the refusal doesn't map to a named category.
7622 
76357623 - `"cyber"`
76367624 
76377625 The request could enable cyber harm, such as malware or exploit development. Benign cybersecurity work can also trigger this category.
from line 7930
79427930 
79437931 - `container: Container or null`
79447932 
7945 Information about the container used in the request (for the code execution tool)
7933 Information about the container used in this request.
79467934 
7935 This will be non-null if a container tool (e.g. code execution) was used.
7936 
79477937 - `id: string`
79487938 
79497939 Identifier for the container used in this request
from line 7960
79707960 
79717961 Skill ID
79727962 
7973 maxLength: 64, minLength: 1
7963 minLength: 1, maxLength: 64
79747964 
79757965 - `version: string`
79767966 
79777967 The resolved version: a skill version ID for custom skills.
79787968 
7979 maxLength: 64, minLength: 1
7969 minLength: 1, maxLength: 64
79807970 
79817971 - `content: array of ContentBlock`
79827972 
from line 8139
81498139 
81508140 - `text: string`
81518141 
8152 minLength: 0
8153 
81548142 - `ThinkingBlock object`
81558143 
81568144 - `type: "thinking"`
from line 8219
82318219 
82328220 For a toolset member tool_use, the toolset family.
82338221 
8234 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
8222 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
82358223 
82368224 - `ServerToolUseBlock object`
82378225 
from line 8667
86798667 
86808668 - `tool_name: string`
86818669 
8682 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
8670 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
86838671 
86848672 - `tool_use_id: string`
86858673 
from line 8685
86978685 
86988686 - `diagnostics: Diagnostics or null`
86998687 
8700 Request-level diagnostics: why the prompt cache could not fully reuse
8701 the prefix of the request named by `diagnostics.previous_message_id`.
8688 Request-level diagnostics. `null` when the request did not supply `diagnostics`, or when it did and no prompt-cache divergence was detected.
87028689 
87038690 - `cache_miss_reason: CacheMissReason or null`
87048691 
from line 8721
87348721 
87358722 Approximate number of input tokens that would have been read from cache had the prefix matched the previous request.
87368723 
8737 - `
8724 - `CacheMissMessagesChanged object`
8725 
8726 - `type

api/messages/batches/create Changed · +125 / -152 lines

from line 31
3131 
3232 List of requests for prompt completion. Each is an individual request to create a Message.
3333 
34 maxItems: 100000, minItems: 1
34 minItems: 1, maxItems: 100000
3535 
3636 - `custom_id: string`
3737 
from line 39
3939 
4040 Must be unique for each request within the Message Batch.
4141 
42 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,64}$
42 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]{1,64}$
4343 
4444 - `params: object`
4545 
from line 159
159159 
160160 - `document_title: string or null`
161161 
162 maxLength: 500, minLength: 1
162 minLength: 1, maxLength: 500
163163 
164164 - `end_char_index: number`
165165 
from line 179
179179 
180180 - `document_title: string or null`
181181 
182 maxLength: 500, minLength: 1
182 minLength: 1, maxLength: 500
183183 
184184 - `end_page_number: number`
185185 
from line 203
203203 
204204 - `document_title: string or null`
205205 
206 maxLength: 500, minLength: 1
206 minLength: 1, maxLength: 500
207207 
208208 - `end_block_index: number`
209209 
from line 227
227227 
228228 - `title: string or null`
229229 
230 maxLength: 512, minLength: 1
230 minLength: 1, maxLength: 512
231231 
232232 - `url: string`
233233 
from line 383
383383 
384384 - `title: optional string or null`
385385 
386 maxLength: 500, minLength: 1
386 minLength: 1, maxLength: 500
387387 
388388 - `SearchResultBlockParam object`
389389 
from line 447
447447 
448448 - `name: string`
449449 
450 maxLength: 200, minLength: 1
450 minLength: 1, maxLength: 200
451451 
452452 - `cache_control: optional CacheControlEphemeral or null`
453453 
from line 483
483483 
484484 For a toolset member tool_use, the toolset family this member belongs to.
485485 
486 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
486 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
487487 
488488 - `ToolResultBlockParam object`
489489 
from line 519
519519 
520520 - `tool_name: string`
521521 
522 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
522 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
523523 
524524 - `cache_control: optional CacheControlEphemeral or null`
525525 
from line 547
547547 
548548 The caller-assigned identifier for this tab, unique within the inventory.
549549 
550 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
550 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
551551 
552552 - `title: string`
553553 
from line 573
573573 
574574 Tabs opened and download state changes during this call. "Nothing to report" is expressed by omitting the field, never by an empty list.
575575 
576 maxItems: 200, minItems: 1
576 minItems: 1, maxItems: 200
577577 
578578 - `BrowserStateChangeTabOpened object`
579579 
from line 591
591591 
592592 The `tab_id` of the opened tab, present in `tabs`.
593593 
594 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
594 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
595595 
596596 - `BrowserStateChangeDownloadStarted object`
597597 
from line 603
603603 
604604 The caller-assigned identifier for this download, stable across the state changes reporting it.
605605 
606 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
606 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
607607 
608608 - `url: string`
609609 
from line 624
624624 
625625 The caller-assigned identifier for this download, stable across the state changes reporting it.
626626 
627 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
627 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
628628 
629629 - `url: string`
630630 
from line 636
636636 
637637 Where the executor saved the file, on the executor's filesystem. Only included when another tool in the same environment can read the file at that path.
638638 
639 pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$, maxLength: 4096
639 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
640640 
641641 - `size_bytes: optional number or null`
642642 
from line 654
654654 
655655 The caller-assigned identifier for this download, stable across the state changes reporting it.
656656 
657 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
657 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
658658 
659659 - `url: string`
660660 
from line 666
666666 
667667 The failure or cancellation detail, when known.
668668 
669 pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$, maxLength: 4096
669 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
670670 
671671 - `is_error: optional boolean`
672672 
from line 674
674674 
675675 For a toolset member tool_result, the toolset family of the paired tool_use.
676676 
677 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
677 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
678678 
679679 - `ServerToolUseBlockParam object`
680680 
from line 1052
10521052 
10531053 - `tool_name: string`
10541054 
1055 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
1055 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
10561056 
10571057 - `cache_control: optional CacheControlEphemeral or null`
10581058 
from line 1093
10931093 
10941094 See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.
10951095 
1096 - `string`
1097 
10961098 - `"claude-fable-5-1" or "claude-opus-5-5" or "claude-mythos-5-1" or 15 more`
10971099 
10981100 The model that will complete your prompt.
from line 1137
11351137 
11361138 Powerful intelligence for long-running agents and coding
11371139 
1138 - `"claude-mythos-preview"`
1139 
1140 New class of intelligence, strongest in coding and cybersecurity
1141 
11421140 - `"claude-opus-4-6"`
11431141 
11441142 Powerful intelligence for long-running agents and coding
from line 1169
11711169 
11721170 High-performance model for agents and coding
11731171 
1174 - `string`
1172 - `"claude-mythos-preview"`
11751173 
1174 **Deprecated**: Will reach end-of-life on June 30, 2026. Please migrate to claude-mythos-5. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.
1175 
1176 New class of intelligence, strongest in coding and cybersecurity
1177 
11761178 - `cache_control: optional CacheControlEphemeral or null`
11771179 
11781180 Top-level cache control automatically applies a cache_control marker to the last cacheable block in the request.
from line 1209
12071209 
12081210 Skill ID
12091211 
1210 maxLength: 64, minLength: 1
1212 minLength: 1, maxLength: 64
12111213 
12121214 - `version: optional string`
12131215 
12141216 Skill version or 'latest' for most recent version
12151217 
1216 maxLength: 64, minLength: 1
1218 minLength: 1, maxLength: 64
12171219 
12181220 - `string`
12191221 
12201222 - `diagnostics: optional DiagnosticsParam or null`
12211223 
1222 Request-level diagnostics. Currently carries the previous response
1223 id for prompt-cache divergence reporting.
1224 Request-level diagnostics. Supply `previous_message_id` to have the response include `diagnostics.cache_miss_reason` explaining any prompt-cache divergence from that prior request.
12241225 
12251226 - `previous_message_id: optional string or null`
12261227 
from line 1251
12501251 
12511252 - `effort: optional "low" or "medium" or "high" or 2 more or null`
12521253 
1253 All possible effort levels.
1254 How much effort the model should put into its response. Higher effort levels may result in more thorough analysis but take longer.
12541255 
1256 Valid values are `low`, `medium`, `high`, `xhigh`, or `max`.
1257 
12551258 - `"low"`
12561259 
12571260 - `"medium"`
from line 1503
15001503 
15011504 This is how the tool will be called by the model and in `tool_use` blocks.
15021505 
1503 maxLength: 128, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,128}$
1506 minLength: 1, maxLength: 128, pattern: ^[a-zA-Z0-9_-]{1,128}$
15041507 
15051508 - `allowed_callers: optional array of "direct" or "code_execution_20250825" or "code_execution_20260120" or "code_execution_20260521"`
15061509 
from line 1720
17171720 
17181721 - `configs: optional BrowserToolsetConfigs or null`
17191722 
1720 Per-member configuration for `browser_toolset_20260801`: one
1721 optional field per member tool, keyed by the member name — the same
1722 name the member's `tool_use` blocks carry. Every member is an
1723 accepted key, and a member's defaults apply wherever its key is
1724 absent. Unknown keys are rejected: the field set is this toolset
1725 version's complete member set.
1723 Sparse per-member overrides, keyed by member name. Absent, null, and {} are equivalent; a member's defaults apply wherever its key is absent.
17261724 
17271725 - `type: optional BrowserTypeConfig or null`
17281726 
from line 2147
21492147 
21502148 - `configs: optional ComputerToolsetConfigs or null`
21512149 
2152 Per-member configuration for `computer_toolset_20260801`: one
2153 optional field per member tool, keyed by the member name — the same
2154 name the member's `tool_use` blocks carry. Every member is an
2155 accepted key, and a member's defaults apply wherever its key is
2156 absent. Unknown keys are rejected: the field set is this toolset
2157 version's complete member set.
2150 Sparse per-member overrides, keyed by member name. Absent, null, and {} are equivalent; a member's defaults apply wherever its key is absent.
21582151 
21592152 - `type: optional ComputerTypeConfig or null`
21602153 
from line 2501
25082501 
25092502 Maximum number of times the tool can be used in the API request.
25102503 
2511 exclusiveMinimum: 0
2504 minimum: 1
25122505 
25132506 - `strict: optional boolean`
25142507 
from line 2517
25242517 
25252518 The city of the user.
25262519 
2527 maxLength: 255, minLength: 1
2520 minLength: 1, maxLength: 255
25282521 
25292522 - `country: optional string or null`
25302523 
25312524 The two letter [ISO country code](https://en.wikipedia.org/wiki/ISO_3166-1_alpha-2) of the user.
25322525 
2533 maxLength: 2, minLength: 2
2526 minLength: 2, maxLength: 2
25342527 
25352528 - `region: optional string or null`
25362529 
25372530 The region of the user.
25382531 
2539 maxLength: 255, minLength: 1
2532 minLength: 1, maxLength: 255
25402533 
25412534 - `timezone: optional string or null`
25422535 
25432536 The [IANA timezone](https://nodatime.org/TimeZones) of the user.
25442537 
2545 maxLength: 255, minLength: 1
2538 minLength: 1, maxLength: 255
25462539 
25472540 - `WebFetchTool20250910 object`
25482541 
from line 2581
25882581 
25892582 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
25902583 
2591 exclusiveMinimum: 0
2584 minimum: 1
25922585 
25932586 - `max_uses: optional number or null`
25942587 
25952588 Maximum number of times the tool can be used in the API request.
25962589 
2597 exclusiveMinimum: 0
2590 minimum: 1
25982591 
25992592 - `strict: optional boolean`
26002593 
from line 2595
26022595 
26032596 - `url_sources: optional WebFetchURLSources or null`
26042597 
2605 Which sources contribute to the set of URLs web fetch may fetch.
2598 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
26062599 
2607 Each key is a tagged variant: `user_input` is `all` or `none`; the
2608 two tool filters are `all`, `none`, `only` (only the named tools'
2609 results) or `except` (every result but the named tools'). A named tool
2610 must be declared in this request's `tools[]`.
2611 
26122600 - `client_tool_results: optional WebFetchURLSourceAll or WebFetchURLSourceNone or WebFetchURLSourceOnly or WebFetchURLSourceExcept`
26132601 
26142602 Which client tools' results contribute fetchable URLs: "all", "none", or an only or except list of client tool names from tools[].
from line 2719
27312719 
27322720 Maximum number of times the tool can be used in the API request.
27332721 
2734 exclusiveMinimum: 0
2722 minimum: 1
27352723 
27362724 - `strict: optional boolean`
27372725 
from line 2773
27852773 
27862774 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
27872775 
2788 exclusiveMinimum: 0
2776 minimum: 1
27892777 
27902778 - `max_uses: optional number or null`
27912779 
27922780 Maximum number of times the tool can be used in the API request.
27932781 
2794 exclusiveMinimum: 0
2782 minimum: 1
27952783 
27962784 - `strict: optional boolean`
27972785 
from line 2787
27992787 
28002788 - `url_sources: optional WebFetchURLSources or null`
28012789 
2802 Which sources contribute to the set of URLs web fetch may fetch.
2790 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
28032791 
2804 Each key is a tagged variant: `user_input` is `all` or `none`; the
2805 two tool filters are `all`, `none`, `only` (only the named tools'
2806 results) or `except` (every result but the named tools'). A named tool
2807 must be declared in this request's `tools[]`.
2808 
28092792 - `WebFetchTool20260309 object`
28102793 
28112794 Web fetch tool with use_cache parameter for bypassing cached content.
from line 2835
28522835 
28532836 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
28542837 
2855 exclusiveMinimum: 0
2838 minimum: 1
28562839 
28572840 - `max_uses: optional number or null`
28582841 
28592842 Maximum number of times the tool can be used in the API request.
28602843 
2861 exclusiveMinimum: 0
2844 minimum: 1
28622845 
28632846 - `strict: optional boolean`
28642847 
from line 2849
28662849 
28672850 - `url_sources: optional WebFetchURLSources or null`
28682851 
2869 Which sources contribute to the set of URLs web fetch may fetch.
2852 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
28702853 
2871 Each key is a tagged variant: `user_input` is `all` or `none`; the
2872 two tool filters are `all`, `none`, `only` (only the named tools'
2873 results) or `except` (every result but the named tools'). A named tool
2874 must be declared in this request's `tools[]`.
2875 
28762854 - `use_cache: optional boolean`
28772855 
28782856 Whether to use cached content. Set to false to bypass the cache and fetch fresh content. Only set to false when the user explicitly requests fresh content or when fetching rapidly-changing sources.
from line 2895
29172895 
29182896 Maximum number of times the tool can be used in the API request.
29192897 
2920 exclusiveMinimum: 0
2898 minimum: 1
29212899 
29222900 - `response_inclusion: optional "full" or "excluded"`
29232901 
from line 2957
29792957 
29802958 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
29812959 
2982 exclusiveMinimum: 0
2960 minimum: 1
29832961 
29842962 - `max_uses: optional number or null`
29852963 
29862964 Maximum number of times the tool can be used in the API request.
29872965 
2988 exclusiveMinimum: 0
2966 minimum: 1
29892967 
29902968 - `response_inclusion: optional "full" or "excluded"`
29912969 
from line 2979
30012979 
30022980 - `url_sources: optional WebFetchURLSources or null`
30032981 
3004 Which sources contribute to the set of URLs web fetch may fetch.
2982 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
30052983 
3006 Each key is a tagged variant: `user_input` is `all` or `none`; the
3007 two tool filters are `all`, `none`, `only` (only the named tools'
3008 results) or `except` (every result but the named tools'). A named tool
3009 must be declared in this request's `tools[]`.
3010 
30112984 - `use_cache: optional boolean`
30122985 
30132986 Whether to use cached content. Set to false to bypass the cache and fetch fresh content. Only set to false when the user explicitly requests fresh content or when fetching rapidly-changing sources.
from line 3067
30943067 
30953068 Note that even with `temperature` of `0.0`, the results will not be fully deterministic.
30963069 
3097 maximum: 1, minimum: 0
3070 minimum: 0, maximum: 1
30983071 
30993072 - `top_k: optional number`
31003073 
from line 3091
31183091 
31193092 Recommended for advanced use cases only.
31203093 
3121 maximum: 1, minimum: 0
3094 minimum: 0, maximum: 1
31223095 
31233096## Returns
31243097 

api/messages/batches/results Changed · +50 / -45 lines

from line 69
6969 
7070 - `container: Container or null`
7171 
72 Information about the container used in the request (for the code execution tool)
72 Information about the container used in this request.
7373 
74 This will be non-null if a container tool (e.g. code execution) was used.
75 
7476 - `id: string`
7577 
7678 Identifier for the container used in this request
from line 99
9799 
98100 Skill ID
99101 
100 maxLength: 64, minLength: 1
102 minLength: 1, maxLength: 64
101103 
102104 - `version: string`
103105 
104106 The resolved version: a skill version ID for custom skills.
105107 
106 maxLength: 64, minLength: 1
108 minLength: 1, maxLength: 64
107109 
108110 - `content: array of ContentBlock`
109111 
from line 278
276278 
277279 - `text: string`
278280 
279 minLength: 0
280 
281281 - `ThinkingBlock object`
282282 
283283 - `type: "thinking"`
from line 358
358358 
359359 For a toolset member tool_use, the toolset family.
360360 
361 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
361 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
362362 
363363 - `ServerToolUseBlock object`
364364 
from line 806
806806 
807807 - `tool_name: string`
808808 
809 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
809 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
810810 
811811 - `tool_use_id: string`
812812 
from line 824
824824 
825825 - `diagnostics: Diagnostics or null`
826826 
827 Request-level diagnostics: why the prompt cache could not fully reuse
828 the prefix of the request named by `diagnostics.previous_message_id`.
827 Request-level diagnostics. `null` when the request did not supply `diagnostics`, or when it did and no prompt-cache divergence was detected.
829828 
830829 - `cache_miss_reason: CacheMissReason or null`
831830 
from line 888
889888 
890889 See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.
891890 
891 - `string`
892 
892893 - `"claude-fable-5-1" or "claude-opus-5-5" or "claude-mythos-5-1" or 15 more`
893894 
894895 The model that will complete your prompt.
from line 932
931932 
932933 Powerful intelligence for long-running agents and coding
933934 
934 - `"claude-mythos-preview"`
935 
936 New class of intelligence, strongest in coding and cybersecurity
937 
938935 - `"claude-opus-4-6"`
939936 
940937 Powerful intelligence for long-running agents and coding
from line 964
967964 
968965 High-performance model for agents and coding
969966 
970 - `string`
967 - `"claude-mythos-preview"`
971968 
969 **Deprecated**: Will reach end-of-life on June 30, 2026. Please migrate to claude-mythos-5. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.
970 
971 New class of intelligence, strongest in coding and cybersecurity
972 
972973 - `role: "assistant"`
973974 
974975 Conversational role of the generated message.
from line 980
979980 
980981 - `stop_details: RefusalStopDetails or null`
981982 
982 Structured information about a refusal.
983 Structured information about why model output stopped.
983984 
985 This is `null` when the `stop_reason` has no additional detail to report.
986 
984987 - `type: "refusal"`
985988 
986989 default: refusal
from line 990
987990 
988991 - `category: "cyber" or "bio" or "frontier_llm" or 2 more or null`
989992 
990 The policy category that triggered a refusal.
993 The policy category that triggered the refusal.
994 
995 `null` when the refusal doesn't map to a named category.
991996 
992997 - `"cyber"`
993998 

api/messages/count_tokens Changed · +79 / -105 lines

from line 127
127127 
128128 - `document_title: string or null`
129129 
130 maxLength: 500, minLength: 1
130 minLength: 1, maxLength: 500
131131 
132132 - `end_char_index: number`
133133 
from line 147
147147 
148148 - `document_title: string or null`
149149 
150 maxLength: 500, minLength: 1
150 minLength: 1, maxLength: 500
151151 
152152 - `end_page_number: number`
153153 
from line 171
171171 
172172 - `document_title: string or null`
173173 
174 maxLength: 500, minLength: 1
174 minLength: 1, maxLength: 500
175175 
176176 - `end_block_index: number`
177177 
from line 195
195195 
196196 - `title: string or null`
197197 
198 maxLength: 512, minLength: 1
198 minLength: 1, maxLength: 512
199199 
200200 - `url: string`
201201 
from line 351
351351 
352352 - `title: optional string or null`
353353 
354 maxLength: 500, minLength: 1
354 minLength: 1, maxLength: 500
355355 
356356 - `SearchResultBlockParam object`
357357 
from line 415
415415 
416416 - `name: string`
417417 
418 maxLength: 200, minLength: 1
418 minLength: 1, maxLength: 200
419419 
420420 - `cache_control: optional CacheControlEphemeral or null`
421421 
from line 451
451451 
452452 For a toolset member tool_use, the toolset family this member belongs to.
453453 
454 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
454 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
455455 
456456 - `ToolResultBlockParam object`
457457 
from line 487
487487 
488488 - `tool_name: string`
489489 
490 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
490 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
491491 
492492 - `cache_control: optional CacheControlEphemeral or null`
493493 
from line 515
515515 
516516 The caller-assigned identifier for this tab, unique within the inventory.
517517 
518 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
518 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
519519 
520520 - `title: string`
521521 
from line 541
541541 
542542 Tabs opened and download state changes during this call. "Nothing to report" is expressed by omitting the field, never by an empty list.
543543 
544 maxItems: 200, minItems: 1
544 minItems: 1, maxItems: 200
545545 
546546 - `BrowserStateChangeTabOpened object`
547547 
from line 559
559559 
560560 The `tab_id` of the opened tab, present in `tabs`.
561561 
562 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
562 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
563563 
564564 - `BrowserStateChangeDownloadStarted object`
565565 
from line 571
571571 
572572 The caller-assigned identifier for this download, stable across the state changes reporting it.
573573 
574 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
574 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
575575 
576576 - `url: string`
577577 
from line 592
592592 
593593 The caller-assigned identifier for this download, stable across the state changes reporting it.
594594 
595 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
595 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
596596 
597597 - `url: string`
598598 
from line 604
604604 
605605 Where the executor saved the file, on the executor's filesystem. Only included when another tool in the same environment can read the file at that path.
606606 
607 pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$, maxLength: 4096
607 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
608608 
609609 - `size_bytes: optional number or null`
610610 
from line 622
622622 
623623 The caller-assigned identifier for this download, stable across the state changes reporting it.
624624 
625 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
625 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
626626 
627627 - `url: string`
628628 
from line 634
634634 
635635 The failure or cancellation detail, when known.
636636 
637 pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$, maxLength: 4096
637 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
638638 
639639 - `is_error: optional boolean`
640640 
from line 642
642642 
643643 For a toolset member tool_result, the toolset family of the paired tool_use.
644644 
645 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
645 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
646646 
647647 - `ServerToolUseBlockParam object`
648648 
from line 1020
10201020 
10211021 - `tool_name: string`
10221022 
1023 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
1023 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
10241024 
10251025 - `cache_control: optional CacheControlEphemeral or null`
10261026 
from line 1061
10611061 
10621062 See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.
10631063 
1064 - `string`
1065 
10641066 - `"claude-fable-5-1" or "claude-opus-5-5" or "claude-mythos-5-1" or 15 more`
10651067 
10661068 The model that will complete your prompt.
from line 1105
11031105 
11041106 Powerful intelligence for long-running agents and coding
11051107 
1106 - `"claude-mythos-preview"`
1107 
1108 New class of intelligence, strongest in coding and cybersecurity
1109 
11101108 - `"claude-opus-4-6"`
11111109 
11121110 Powerful intelligence for long-running agents and coding
from line 1137
11391137 
11401138 High-performance model for agents and coding
11411139 
1142 - `string`
1140 - `"claude-mythos-preview"`
11431141 
1142 **Deprecated**: Will reach end-of-life on June 30, 2026. Please migrate to claude-mythos-5. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.
1143 
1144 New class of intelligence, strongest in coding and cybersecurity
1145 
11441146- `cache_control: optional CacheControlEphemeral or null`
11451147 
11461148 Top-level cache control automatically applies a cache_control marker to the last cacheable block in the request.
from line 1153
11511153 
11521154 - `effort: optional "low" or "medium" or "high" or 2 more or null`
11531155 
1154 All possible effort levels.
1156 How much effort the model should put into its response. Higher effort levels may result in more thorough analysis but take longer.
11551157 
1158 Valid values are `low`, `medium`, `high`, `xhigh`, or `max`.
1159 
11561160 - `"low"`
11571161 
11581162 - `"medium"`
from line 1381
13771381 
13781382 This is how the tool will be called by the model and in `tool_use` blocks.
13791383 
1380 maxLength: 128, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,128}$
1384 minLength: 1, maxLength: 128, pattern: ^[a-zA-Z0-9_-]{1,128}$
13811385 
13821386 - `allowed_callers: optional array of "direct" or "code_execution_20250825" or "code_execution_20260120" or "code_execution_20260521"`
13831387 
from line 1598
15941598 
15951599 - `configs: optional BrowserToolsetConfigs or null`
15961600 
1597 Per-member configuration for `browser_toolset_20260801`: one
1598 optional field per member tool, keyed by the member name — the same
1599 name the member's `tool_use` blocks carry. Every member is an
1600 accepted key, and a member's defaults apply wherever its key is
1601 absent. Unknown keys are rejected: the field set is this toolset
1602 version's complete member set.
1601 Sparse per-member overrides, keyed by member name. Absent, null, and {} are equivalent; a member's defaults apply wherever its key is absent.
16031602 
16041603 - `type: optional BrowserTypeConfig or null`
16051604 
from line 2025
20262025 
20272026 - `configs: optional ComputerToolsetConfigs or null`
20282027 
2029 Per-member configuration for `computer_toolset_20260801`: one
2030 optional field per member tool, keyed by the member name — the same
2031 name the member's `tool_use` blocks carry. Every member is an
2032 accepted key, and a member's defaults apply wherever its key is
2033 absent. Unknown keys are rejected: the field set is this toolset
2034 version's complete member set.
2028 Sparse per-member overrides, keyed by member name. Absent, null, and {} are equivalent; a member's defaults apply wherever its key is absent.
20352029 
20362030 - `type: optional ComputerTypeConfig or null`
20372031 
from line 2379
23852379 
23862380 Maximum number of times the tool can be used in the API request.
23872381 
2388 exclusiveMinimum: 0
2382 minimum: 1
23892383 
23902384 - `strict: optional boolean`
23912385 
from line 2395
24012395 
24022396 The city of the user.
24032397 
2404 maxLength: 255, minLength: 1
2398 minLength: 1, maxLength: 255
24052399 
24062400 - `country: optional string or null`
24072401 
24082402 The two letter [ISO country code](https://en.wikipedia.org/wiki/ISO_3166-1_alpha-2) of the user.
24092403 
2410 maxLength: 2, minLength: 2
2404 minLength: 2, maxLength: 2
24112405 
24122406 - `region: optional string or null`
24132407 
24142408 The region of the user.
24152409 
2416 maxLength: 255, minLength: 1
2410 minLength: 1, maxLength: 255
24172411 
24182412 - `timezone: optional string or null`
24192413 
24202414 The [IANA timezone](https://nodatime.org/TimeZones) of the user.
24212415 
2422 maxLength: 255, minLength: 1
2416 minLength: 1, maxLength: 255
24232417 
24242418 - `WebFetchTool20250910 object`
24252419 
from line 2459
24652459 
24662460 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
24672461 
2468 exclusiveMinimum: 0
2462 minimum: 1
24692463 
24702464 - `max_uses: optional number or null`
24712465 
24722466 Maximum number of times the tool can be used in the API request.
24732467 
2474 exclusiveMinimum: 0
2468 minimum: 1
24752469 
24762470 - `strict: optional boolean`
24772471 
from line 2473
24792473 
24802474 - `url_sources: optional WebFetchURLSources or null`
24812475 
2482 Which sources contribute to the set of URLs web fetch may fetch.
2476 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
24832477 
2484 Each key is a tagged variant: `user_input` is `all` or `none`; the
2485 two tool filters are `all`, `none`, `only` (only the named tools'
2486 results) or `except` (every result but the named tools'). A named tool
2487 must be declared in this request's `tools[]`.
2488 
24892478 - `client_tool_results: optional WebFetchURLSourceAll or WebFetchURLSourceNone or WebFetchURLSourceOnly or WebFetchURLSourceExcept`
24902479 
24912480 Which client tools' results contribute fetchable URLs: "all", "none", or an only or except list of client tool names from tools[].
from line 2597
26082597 
26092598 Maximum number of times the tool can be used in the API request.
26102599 
2611 exclusiveMinimum: 0
2600 minimum: 1
26122601 
26132602 - `strict: optional boolean`
26142603 
from line 2651
26622651 
26632652 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
26642653 
2665 exclusiveMinimum: 0
2654 minimum: 1
26662655 
26672656 - `max_uses: optional number or null`
26682657 
26692658 Maximum number of times the tool can be used in the API request.
26702659 
2671 exclusiveMinimum: 0
2660 minimum: 1
26722661 
26732662 - `strict: optional boolean`
26742663 
from line 2665
26762665 
26772666 - `url_sources: optional WebFetchURLSources or null`
26782667 
2679 Which sources contribute to the set of URLs web fetch may fetch.
2668 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
26802669 
2681 Each key is a tagged variant: `user_input` is `all` or `none`; the
2682 two tool filters are `all`, `none`, `only` (only the named tools'
2683 results) or `except` (every result but the named tools'). A named tool
2684 must be declared in this request's `tools[]`.
2685 
26862670 - `WebFetchTool20260309 object`
26872671 
26882672 Web fetch tool with use_cache parameter for bypassing cached content.
from line 2713
27292713 
27302714 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
27312715 
2732 exclusiveMinimum: 0
2716 minimum: 1
27332717 
27342718 - `max_uses: optional number or null`
27352719 
27362720 Maximum number of times the tool can be used in the API request.
27372721 
2738 exclusiveMinimum: 0
2722 minimum: 1
27392723 
27402724 - `strict: optional boolean`
27412725 
from line 2727
27432727 
27442728 - `url_sources: optional WebFetchURLSources or null`
27452729 
2746 Which sources contribute to the set of URLs web fetch may fetch.
2730 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
27472731 
2748 Each key is a tagged variant: `user_input` is `all` or `none`; the
2749 two tool filters are `all`, `none`, `only` (only the named tools'
2750 results) or `except` (every result but the named tools'). A named tool
2751 must be declared in this request's `tools[]`.
2752 
27532732 - `use_cache: optional boolean`
27542733 
27552734 Whether to use cached content. Set to false to bypass the cache and fetch fresh content. Only set to false when the user explicitly requests fresh content or when fetching rapidly-changing sources.
from line 2773
27942773 
27952774 Maximum number of times the tool can be used in the API request.
27962775 
2797 exclusiveMinimum: 0
2776 minimum: 1
27982777 
27992778 - `response_inclusion: optional "full" or "excluded"`
28002779 
from line 2835
28562835 
28572836 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
28582837 
2859 exclusiveMinimum: 0
2838 minimum: 1
28602839 
28612840 - `max_uses: optional number or null`
28622841 
28632842 Maximum number of times the tool can be used in the API request.
28642843 
2865 exclusiveMinimum: 0
2844 minimum: 1
28662845 
28672846 - `response_inclusion: optional "full" or "excluded"`
28682847 
from line 2857
28782857 
28792858 - `url_sources: optional WebFetchURLSources or null`
28802859 
2881 Which sources contribute to the set of URLs web fetch may fetch.
2882 
2883 Each key is a tagged variant: `user_input` is `all` or `none`; the
2884 two tool filters are `all`, `none`, `only` (only the named tools'
2885 results) or `except` (every result but the named tools'). A named tool
2886 must be declared in this request's `tools[]`.
2860 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
28872861 
28882862 - `use_cache: optional boolean`
28892863 

api/messages/create Changed · +179 / -197 lines

from line 139
139139 
140140 - `document_title: string or null`
141141 
142 maxLength: 500, minLength: 1
142 minLength: 1, maxLength: 500
143143 
144144 - `end_char_index: number`
145145 
from line 159
159159 
160160 - `document_title: string or null`
161161 
162 maxLength: 500, minLength: 1
162 minLength: 1, maxLength: 500
163163 
164164 - `end_page_number: number`
165165 
from line 183
183183 
184184 - `document_title: string or null`
185185 
186 maxLength: 500, minLength: 1
186 minLength: 1, maxLength: 500
187187 
188188 - `end_block_index: number`
189189 
from line 207
207207 
208208 - `title: string or null`
209209 
210 maxLength: 512, minLength: 1
210 minLength: 1, maxLength: 512
211211 
212212 - `url: string`
213213 
from line 363
363363 
364364 - `title: optional string or null`
365365 
366 maxLength: 500, minLength: 1
366 minLength: 1, maxLength: 500
367367 
368368 - `SearchResultBlockParam object`
369369 
from line 427
427427 
428428 - `name: string`
429429 
430 maxLength: 200, minLength: 1
430 minLength: 1, maxLength: 200
431431 
432432 - `cache_control: optional CacheControlEphemeral or null`
433433 
from line 463
463463 
464464 For a toolset member tool_use, the toolset family this member belongs to.
465465 
466 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
466 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
467467 
468468 - `ToolResultBlockParam object`
469469 
from line 499
499499 
500500 - `tool_name: string`
501501 
502 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
502 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
503503 
504504 - `cache_control: optional CacheControlEphemeral or null`
505505 
from line 527
527527 
528528 The caller-assigned identifier for this tab, unique within the inventory.
529529 
530 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
530 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
531531 
532532 - `title: string`
533533 
from line 553
553553 
554554 Tabs opened and download state changes during this call. "Nothing to report" is expressed by omitting the field, never by an empty list.
555555 
556 maxItems: 200, minItems: 1
556 minItems: 1, maxItems: 200
557557 
558558 - `BrowserStateChangeTabOpened object`
559559 
from line 571
571571 
572572 The `tab_id` of the opened tab, present in `tabs`.
573573 
574 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
574 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
575575 
576576 - `BrowserStateChangeDownloadStarted object`
577577 
from line 583
583583 
584584 The caller-assigned identifier for this download, stable across the state changes reporting it.
585585 
586 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
586 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
587587 
588588 - `url: string`
589589 
from line 604
604604 
605605 The caller-assigned identifier for this download, stable across the state changes reporting it.
606606 
607 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
607 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
608608 
609609 - `url: string`
610610 
from line 616
616616 
617617 Where the executor saved the file, on the executor's filesystem. Only included when another tool in the same environment can read the file at that path.
618618 
619 pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$, maxLength: 4096
619 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
620620 
621621 - `size_bytes: optional number or null`
622622 
from line 634
634634 
635635 The caller-assigned identifier for this download, stable across the state changes reporting it.
636636 
637 maxLength: 4096, minLength: 1, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
637 minLength: 1, maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
638638 
639639 - `url: string`
640640 
from line 646
646646 
647647 The failure or cancellation detail, when known.
648648 
649 pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$, maxLength: 4096
649 maxLength: 4096, pattern: ^[^\x00-\x1f\x7f-\x9f\u2028\u2029]*$
650650 
651651 - `is_error: optional boolean`
652652 
from line 654
654654 
655655 For a toolset member tool_result, the toolset family of the paired tool_use.
656656 
657 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
657 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
658658 
659659 - `ServerToolUseBlockParam object`
660660 
from line 1032
10321032 
10331033 - `tool_name: string`
10341034 
1035 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
1035 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
10361036 
10371037 - `cache_control: optional CacheControlEphemeral or null`
10381038 
from line 1073
10731073 
10741074 See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.
10751075 
1076 - `string`
1077 
10761078 - `"claude-fable-5-1" or "claude-opus-5-5" or "claude-mythos-5-1" or 15 more`
10771079 
10781080 The model that will complete your prompt.
from line 1117
11151117 
11161118 Powerful intelligence for long-running agents and coding
11171119 
1118 - `"claude-mythos-preview"`
1119 
1120 New class of intelligence, strongest in coding and cybersecurity
1121 
11221120 - `"claude-opus-4-6"`
11231121 
11241122 Powerful intelligence for long-running agents and coding
from line 1149
11511149 
11521150 High-performance model for agents and coding
11531151 
1154 - `string`
1152 - `"claude-mythos-preview"`
11551153 
1154 **Deprecated**: Will reach end-of-life on June 30, 2026. Please migrate to claude-mythos-5. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.
1155 
1156 New class of intelligence, strongest in coding and cybersecurity
1157 
11561158- `cache_control: optional CacheControlEphemeral or null`
11571159 
11581160 Top-level cache control automatically applies a cache_control marker to the last cacheable block in the request.
from line 1189
11871189 
11881190 Skill ID
11891191 
1190 maxLength: 64, minLength: 1
1192 minLength: 1, maxLength: 64
11911193 
11921194 - `version: optional string`
11931195 
11941196 Skill version or 'latest' for most recent version
11951197 
1196 maxLength: 64, minLength: 1
1198 minLength: 1, maxLength: 64
11971199 
11981200 - `string`
11991201 
12001202- `diagnostics: optional DiagnosticsParam or null`
12011203 
1202 Request-level diagnostics. Currently carries the previous response
1203 id for prompt-cache divergence reporting.
1204 Request-level diagnostics. Supply `previous_message_id` to have the response include `diagnostics.cache_miss_reason` explaining any prompt-cache divergence from that prior request.
12041205 
12051206 - `previous_message_id: optional string or null`
12061207 
from line 1231
12301231 
12311232 - `effort: optional "low" or "medium" or "high" or 2 more or null`
12321233 
1233 All possible effort levels.
1234 How much effort the model should put into its response. Higher effort levels may result in more thorough analysis but take longer.
12341235 
1236 Valid values are `low`, `medium`, `high`, `xhigh`, or `max`.
1237 
12351238 - `"low"`
12361239 
12371240 - `"medium"`
from line 1483
14801483 
14811484 This is how the tool will be called by the model and in `tool_use` blocks.
14821485 
1483 maxLength: 128, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,128}$
1486 minLength: 1, maxLength: 128, pattern: ^[a-zA-Z0-9_-]{1,128}$
14841487 
14851488 - `allowed_callers: optional array of "direct" or "code_execution_20250825" or "code_execution_20260120" or "code_execution_20260521"`
14861489 
from line 1700
16971700 
16981701 - `configs: optional BrowserToolsetConfigs or null`
16991702 
1700 Per-member configuration for `browser_toolset_20260801`: one
1701 optional field per member tool, keyed by the member name — the same
1702 name the member's `tool_use` blocks carry. Every member is an
1703 accepted key, and a member's defaults apply wherever its key is
1704 absent. Unknown keys are rejected: the field set is this toolset
1705 version's complete member set.
1703 Sparse per-member overrides, keyed by member name. Absent, null, and {} are equivalent; a member's defaults apply wherever its key is absent.
17061704 
17071705 - `type: optional BrowserTypeConfig or null`
17081706 
from line 2127
21292127 
21302128 - `configs: optional ComputerToolsetConfigs or null`
21312129 
2132 Per-member configuration for `computer_toolset_20260801`: one
2133 optional field per member tool, keyed by the member name — the same
2134 name the member's `tool_use` blocks carry. Every member is an
2135 accepted key, and a member's defaults apply wherever its key is
2136 absent. Unknown keys are rejected: the field set is this toolset
2137 version's complete member set.
2130 Sparse per-member overrides, keyed by member name. Absent, null, and {} are equivalent; a member's defaults apply wherever its key is absent.
21382131 
21392132 - `type: optional ComputerTypeConfig or null`
21402133 
from line 2481
24882481 
24892482 Maximum number of times the tool can be used in the API request.
24902483 
2491 exclusiveMinimum: 0
2484 minimum: 1
24922485 
24932486 - `strict: optional boolean`
24942487 
from line 2497
25042497 
25052498 The city of the user.
25062499 
2507 maxLength: 255, minLength: 1
2500 minLength: 1, maxLength: 255
25082501 
25092502 - `country: optional string or null`
25102503 
25112504 The two letter [ISO country code](https://en.wikipedia.org/wiki/ISO_3166-1_alpha-2) of the user.
25122505 
2513 maxLength: 2, minLength: 2
2506 minLength: 2, maxLength: 2
25142507 
25152508 - `region: optional string or null`
25162509 
25172510 The region of the user.
25182511 
2519 maxLength: 255, minLength: 1
2512 minLength: 1, maxLength: 255
25202513 
25212514 - `timezone: optional string or null`
25222515 
25232516 The [IANA timezone](https://nodatime.org/TimeZones) of the user.
25242517 
2525 maxLength: 255, minLength: 1
2518 minLength: 1, maxLength: 255
25262519 
25272520 - `WebFetchTool20250910 object`
25282521 
from line 2561
25682561 
25692562 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
25702563 
2571 exclusiveMinimum: 0
2564 minimum: 1
25722565 
25732566 - `max_uses: optional number or null`
25742567 
25752568 Maximum number of times the tool can be used in the API request.
25762569 
2577 exclusiveMinimum: 0
2570 minimum: 1
25782571 
25792572 - `strict: optional boolean`
25802573 
from line 2575
25822575 
25832576 - `url_sources: optional WebFetchURLSources or null`
25842577 
2585 Which sources contribute to the set of URLs web fetch may fetch.
2578 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
25862579 
2587 Each key is a tagged variant: `user_input` is `all` or `none`; the
2588 two tool filters are `all`, `none`, `only` (only the named tools'
2589 results) or `except` (every result but the named tools'). A named tool
2590 must be declared in this request's `tools[]`.
2591 
25922580 - `client_tool_results: optional WebFetchURLSourceAll or WebFetchURLSourceNone or WebFetchURLSourceOnly or WebFetchURLSourceExcept`
25932581 
25942582 Which client tools' results contribute fetchable URLs: "all", "none", or an only or except list of client tool names from tools[].
from line 2699
27112699 
27122700 Maximum number of times the tool can be used in the API request.
27132701 
2714 exclusiveMinimum: 0
2702 minimum: 1
27152703 
27162704 - `strict: optional boolean`
27172705 
from line 2753
27652753 
27662754 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
27672755 
2768 exclusiveMinimum: 0
2756 minimum: 1
27692757 
27702758 - `max_uses: optional number or null`
27712759 
27722760 Maximum number of times the tool can be used in the API request.
27732761 
2774 exclusiveMinimum: 0
2762 minimum: 1
27752763 
27762764 - `strict: optional boolean`
27772765 
from line 2767
27792767 
27802768 - `url_sources: optional WebFetchURLSources or null`
27812769 
2782 Which sources contribute to the set of URLs web fetch may fetch.
2770 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
27832771 
2784 Each key is a tagged variant: `user_input` is `all` or `none`; the
2785 two tool filters are `all`, `none`, `only` (only the named tools'
2786 results) or `except` (every result but the named tools'). A named tool
2787 must be declared in this request's `tools[]`.
2788 
27892772 - `WebFetchTool20260309 object`
27902773 
27912774 Web fetch tool with use_cache parameter for bypassing cached content.
from line 2815
28322815 
28332816 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
28342817 
2835 exclusiveMinimum: 0
2818 minimum: 1
28362819 
28372820 - `max_uses: optional number or null`
28382821 
28392822 Maximum number of times the tool can be used in the API request.
28402823 
2841 exclusiveMinimum: 0
2824 minimum: 1
28422825 
28432826 - `strict: optional boolean`
28442827 
from line 2829
28462829 
28472830 - `url_sources: optional WebFetchURLSources or null`
28482831 
2849 Which sources contribute to the set of URLs web fetch may fetch.
2832 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
28502833 
2851 Each key is a tagged variant: `user_input` is `all` or `none`; the
2852 two tool filters are `all`, `none`, `only` (only the named tools'
2853 results) or `except` (every result but the named tools'). A named tool
2854 must be declared in this request's `tools[]`.
2855 
28562834 - `use_cache: optional boolean`
28572835 
28582836 Whether to use cached content. Set to false to bypass the cache and fetch fresh content. Only set to false when the user explicitly requests fresh content or when fetching rapidly-changing sources.
from line 2875
28972875 
28982876 Maximum number of times the tool can be used in the API request.
28992877 
2900 exclusiveMinimum: 0
2878 minimum: 1
29012879 
29022880 - `response_inclusion: optional "full" or "excluded"`
29032881 
from line 2937
29592937 
29602938 Maximum number of tokens used by including web page text content in the context. The limit is approximate and does not apply to binary content such as PDFs.
29612939 
2962 exclusiveMinimum: 0
2940 minimum: 1
29632941 
29642942 - `max_uses: optional number or null`
29652943 
29662944 Maximum number of times the tool can be used in the API request.
29672945 
2968 exclusiveMinimum: 0
2946 minimum: 1
29692947 
29702948 - `response_inclusion: optional "full" or "excluded"`
29712949 
from line 2959
29812959 
29822960 - `url_sources: optional WebFetchURLSources or null`
29832961 
2984 Which sources contribute to the set of URLs web fetch may fetch.
2962 Which sources contribute to the set of URLs the tool may fetch. Omitted means every source.
29852963 
2986 Each key is a tagged variant: `user_input` is `all` or `none`; the
2987 two tool filters are `all`, `none`, `only` (only the named tools'
2988 results) or `except` (every result but the named tools'). A named tool
2989 must be declared in this request's `tools[]`.
2990 
29912964 - `use_cache: optional boolean`
29922965 
29932966 Whether to use cached content. Set to false to bypass the cache and fetch fresh content. Only set to false when the user explicitly requests fresh content or when fetching rapidly-changing sources.
from line 3047
30743047 
30753048 Note that even with `temperature` of `0.0`, the results will not be fully deterministic.
30763049 
3077 maximum: 1, minimum: 0
3050 minimum: 0, maximum: 1
30783051 
30793052- `top_k: optional number`
30803053 
from line 3071
30983071 
30993072 Recommended for advanced use cases only.
31003073 
3101 maximum: 1, minimum: 0
3074 minimum: 0, maximum: 1
31023075 
31033076## Returns
31043077 
from line 3093
31203093 
31213094 - `container: Container or null`
31223095 
3123 Information about the container used in the request (for the code execution tool)
3096 Information about the container used in this request.
31243097 
3098 This will be non-null if a container tool (e.g. code execution) was used.
3099 
31253100 - `id: string`
31263101 
31273102 Identifier for the container used in this request
from line 3123
31483123 
31493124 Skill ID
31503125 
3151 maxLength: 64, minLength: 1
3126 minLength: 1, maxLength: 64
31523127 
31533128 - `version: string`
31543129 
31553130 The resolved version: a skill version ID for custom skills.
31563131 
3157 maxLength: 64, minLength: 1
3132 minLength: 1, maxLength: 64
31583133 
31593134 - `content: array of ContentBlock`
31603135 
from line 3302
33273302 
33283303 - `text: string`
33293304 
3330 minLength: 0
3331 
33323305 - `ThinkingBlock object`
33333306 
33343307 - `type: "thinking"`
from line 3382
34093382 
34103383 For a toolset member tool_use, the toolset family.
34113384 
3412 maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$
3385 minLength: 1, maxLength: 64, pattern: ^[a-zA-Z0-9_-]+$
34133386 
34143387 - `ServerToolUseBlock object`
34153388 
from line 3830
38573830 
38583831 - `tool_name: string`
38593832 
3860 maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$
3833 minLength: 1, maxLength: 256, pattern: ^[a-zA-Z0-9_-]{1,256}$
38613834 
38623835 - `tool_use_id: string`
38633836 
from line 3848
38753848 
38763849 - `diagnostics: Diagnostics or null`
38773850 
3878 Request-level diagnostics: why the prompt cache could not fully reuse
3879 the prefix of the request named by `diagnostics.previous_message_id`.
3851 Request-level diagnostics. `null` when the request did not supply `diagnostics`, or when it did and no prompt-cache divergence was detected.
38803852 
38813853 - `cache_miss_reason: CacheMissReason or null`
38823854 
from line 3912
39403912 
39413913 See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.
39423914 
3915 - `string`
3916 
39433917 - `"claude-fable-5-1" or "claude-opus-5-5" or "claude-mythos-5-1" or 15 more`
39443918 
39453919 The model that will complete your prompt.
from line 3956
39823956 
39833957 Powerful intelligence for long-running agents and coding
39843958 
3985 - `"claude-mythos-preview"`
3986 
3987 New class of intelligence, strongest in coding and cybersecurity
3988 
39893959 - `"claude-opus-4-6"`
39903960 
39913961 Powerful intelligence for long-running agents and coding
from line 3988
40183988 
40193989 High-performance model for agents and coding
40203990 
4021 - `string`
3991 - `"claude-mythos-preview"`
40223992 
3993 **Deprecated**: Will reach end-of-life on June 30, 2026. Please migrate to claude-mythos-5. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.
3994 
3995 New class of intelligence, strongest in coding and cybersecurity
3996 
40233997 - `role: "assistant"`
40243998 
40253999 Conversational role of the generated message.
from line 4004
40304004 
40314005 - `stop_details: RefusalStopDetails or null`
40324006 
4033 Structured information about a refusal.
4007 Structured information about why model output stopped.
40344008 
4009 This is `null` when the `stop_reason` has no additional detail to report.
4010 
40354011 - `type: "refusal"`
40364012 
40374013 default: refusal
from line 4014
40384014 
40394015 - `category: "cyber" or "bio" or "frontier_llm" or 2 more or null`
40404016 
4041 The policy category that triggered a refusal.
4017 The policy category that triggered the refusal.
40424018 
4019 `null` when the refusal doesn't map to a named category.
4020 
40434021 - `"cyber"`
40444022 
40454023 The request could enable cyber harm, such as malware or exploit development. Benign cybersecurity work can also trigger this category.
from line 4204
42264204 
42274205 - `container: Container or null`
42284206 
4229 Information about the container used in the request (for the code execution tool)
4207 Information about the container used in this request.
42304208 
4209 This will be non-null if a container tool (e.g. code execution) was used.
4210 
42314211 - `stop_details: RefusalStopDetails or null`
42324212 
4233 Structured information about a refusal.
4213 Structured information about why model output stopped.
4214 
4215 This is `null` when the `stop_reason` has no additional detail to report.
42344216 
42354217 - `stop_reason: StopReason or null`
42364218 

api/models Changed · +24 / -24 lines

from line 29
2929 
3030 Defaults to `20`. Ranges from `1` to `1000`.
3131 
32 default: 20, maximum: 1000, minimum: 1
32 default: 20, minimum: 1, maximum: 1000
3333 
3434### Headers
3535 
from line 163
163163 
164164 - `capabilities: ModelCapabilities or null`
165165 
166 Model capability information.
166 Object mapping capability names to their support details. Keys are always present for all known capabilities.
167167 
168168 - `batch: CapabilitySupport`
169169 
from line 187
187187 
188188 - `clear_thinking_20251015: CapabilitySupport or null`
189189 
190 Indicates whether a capability is supported.
190 Whether the clear_thinking_20251015 strategy is supported.
191191 
192192 - `clear_tool_uses_20250919: CapabilitySupport or null`
193193 
194 Indicates whether a capability is supported.
194 Whether the clear_tool_uses_20250919 strategy is supported.
195195 
196196 - `compact_20260112: CapabilitySupport or null`
197197 
198 Indicates whether a capability is supported.
198 Whether the compact_20260112 strategy is supported.
199199 
200200 - `supported: boolean`
201201 
from line 227
227227 
228228 - `xhigh: CapabilitySupport or null`
229229 
230 Indicates whether a capability is supported.
230 Whether the model supports xhigh effort level.
231231 
232232 - `image_input: CapabilitySupport`
233233 
from line 526
526526 
527527 - `capabilities: ModelCapabilities or null`
528528 
529 Model capability information.
529 Object mapping capability names to their support details. Keys are always present for all known capabilities.
530530 
531531 - `batch: CapabilitySupport`
532532 
from line 550
550550 
551551 - `clear_thinking_20251015: CapabilitySupport or null`
552552 
553 Indicates whether a capability is supported.
553 Whether the clear_thinking_20251015 strategy is supported.
554554 
555555 - `clear_tool_uses_20250919: CapabilitySupport or null`
556556 
557 Indicates whether a capability is supported.
557 Whether the clear_tool_uses_20250919 strategy is supported.
558558 
559559 - `compact_20260112: CapabilitySupport or null`
560560 
561 Indicates whether a capability is supported.
561 Whether the compact_20260112 strategy is supported.
562562 
563563 - `supported: boolean`
564564 
from line 590
590590 
591591 - `xhigh: CapabilitySupport or null`
592592 
593 Indicates whether a capability is supported.
593 Whether the model supports xhigh effort level.
594594 
595595 - `image_input: CapabilitySupport`
596596 
from line 744
744744 
745745 - `clear_thinking_20251015: CapabilitySupport or null`
746746 
747 Indicates whether a capability is supported.
747 Whether the clear_thinking_20251015 strategy is supported.
748748 
749749 - `supported: boolean`
750750 
from line 752
752752 
753753 - `clear_tool_uses_20250919: CapabilitySupport or null`
754754 
755 Indicates whether a capability is supported.
755 Whether the clear_tool_uses_20250919 strategy is supported.
756756 
757757 - `compact_20260112: CapabilitySupport or null`
758758 
759 Indicates whether a capability is supported.
759 Whether the compact_20260112 strategy is supported.
760760 
761761 - `supported: boolean`
762762 
from line 794
794794 
795795 - `xhigh: CapabilitySupport or null`
796796 
797 Indicates whether a capability is supported.
797 Whether the model supports xhigh effort level.
798798 
799799### Model Capabilities
800800 
from line 824
824824 
825825 - `clear_thinking_20251015: CapabilitySupport or null`
826826 
827 Indicates whether a capability is supported.
827 Whether the clear_thinking_20251015 strategy is supported.
828828 
829829 - `clear_tool_uses_20250919: CapabilitySupport or null`
830830 
831 Indicates whether a capability is supported.
831 Whether the clear_tool_uses_20250919 strategy is supported.
832832 
833833 - `compact_20260112: CapabilitySupport or null`
834834 
835 Indicates whether a capability is supported.
835 Whether the compact_20260112 strategy is supported.
836836 
837837 - `supported: boolean`
838838 
from line 864
864864 
865865 - `xhigh: CapabilitySupport or null`
866866 
867 Indicates whether a capability is supported.
867 Whether the model supports xhigh effort level.
868868 
869869 - `image_input: CapabilitySupport`
870870 
from line 916
916916 
917917 - `capabilities: ModelCapabilities or null`
918918 
919 Model capability information.
919 Object mapping capability names to their support details. Keys are always present for all known capabilities.
920920 
921921 - `batch: CapabilitySupport`
922922 
from line 940
940940 
941941 - `clear_thinking_20251015: CapabilitySupport or null`
942942 
943 Indicates whether a capability is supported.
943 Whether the clear_thinking_20251015 strategy is supported.
944944 
945945 - `clear_tool_uses_20250919: CapabilitySupport or null`
946946 
947 Indicates whether a capability is supported.
947 Whether the clear_tool_uses_20250919 strategy is supported.
948948 
949949 - `compact_20260112: CapabilitySupport or null`
950950 
951 Indicates whether a capability is supported.
951 Whether the compact_20260112 strategy is supported.
952952 
953953 - `supported: boolean`
954954 
from line 980
980980 
981981 - `xhigh: CapabilitySupport or null`
982982 
983 Indicates whether a capability is supported.
983 Whether the model supports xhigh effort level.
984984 
985985 - `image_input: CapabilitySupport`
986986 

api/models/list Changed · +6 / -6 lines

from line 27
2727 
2828 Defaults to `20`. Ranges from `1` to `1000`.
2929 
30 default: 20, maximum: 1000, minimum: 1
30 default: 20, minimum: 1, maximum: 1000
3131 
3232## Headers
3333 
from line 161
161161 
162162 - `capabilities: ModelCapabilities or null`
163163 
164 Model capability information.
164 Object mapping capability names to their support details. Keys are always present for all known capabilities.
165165 
166166 - `batch: CapabilitySupport`
167167 
from line 185
185185 
186186 - `clear_thinking_20251015: CapabilitySupport or null`
187187 
188 Indicates whether a capability is supported.
188 Whether the clear_thinking_20251015 strategy is supported.
189189 
190190 - `clear_tool_uses_20250919: CapabilitySupport or null`
191191 
192 Indicates whether a capability is supported.
192 Whether the clear_tool_uses_20250919 strategy is supported.
193193 
194194 - `compact_20260112: CapabilitySupport or null`
195195 
196 Indicates whether a capability is supported.
196 Whether the compact_20260112 strategy is supported.
197197 
198198 - `supported: boolean`
199199 
from line 225
225225 
226226 - `xhigh: CapabilitySupport or null`
227227 
228 Indicates whether a capability is supported.
228 Whether the model supports xhigh effort level.
229229 
230230 - `image_input: CapabilitySupport`
231231 

api/models/retrieve Changed · +5 / -5 lines

from line 149
149149 
150150 - `capabilities: ModelCapabilities or null`
151151 
152 Model capability information.
152 Object mapping capability names to their support details. Keys are always present for all known capabilities.
153153 
154154 - `batch: CapabilitySupport`
155155 
from line 173
173173 
174174 - `clear_thinking_20251015: CapabilitySupport or null`
175175 
176 Indicates whether a capability is supported.
176 Whether the clear_thinking_20251015 strategy is supported.
177177 
178178 - `clear_tool_uses_20250919: CapabilitySupport or null`
179179 
180 Indicates whether a capability is supported.
180 Whether the clear_tool_uses_20250919 strategy is supported.
181181 
182182 - `compact_20260112: CapabilitySupport or null`
183183 
184 Indicates whether a capability is supported.
184 Whether the compact_20260112 strategy is supported.
185185 
186186 - `supported: boolean`
187187 
from line 213
213213 
214214 - `xhigh: CapabilitySupport or null`
215215 
216 Indicates whether a capability is supported.
216 Whether the model supports xhigh effort level.
217217 
218218 - `image_input: CapabilitySupport`
219219 

build-with-claude/batch-processing Changed · +3 / -3 lines

from line 446
446446 
447447### Tracking your batch
448448 
449The Message Batch's `processing_status` field indicates the stage of processing the batch is in. It starts as `in_progress`, then updates to `ended` once all the requests in the batch have finished processing, and results are ready. You can monitor the state of your batch by visiting the [Console](https://platform.claude.com/settings/workspaces/default/batches), or using the [retrieval endpoint](https://platform.claude.com/docs/en/api/retrieving-message-batches).
449The Message Batch's `processing_status` field indicates the stage of processing the batch is in. It starts as `in_progress`, then updates to `ended` once all the requests in the batch have finished processing, and results are ready. You can monitor the state of your batch by visiting the [Console](https://platform.claude.com/settings/workspaces/default/batches), or using the [retrieval endpoint](https://platform.claude.com/docs/en/api/messages/batches/retrieve).
450450 
451451#### Polling for Message Batch completion
452452 
from line 604
604604 
605605### Listing all Message Batches
606606 
607You can list all Message Batches in your Workspace using the [list endpoint](https://platform.claude.com/docs/en/api/listing-message-batches). The API supports pagination, automatically fetching additional pages as needed:
607You can list all Message Batches in your Workspace using the [list endpoint](https://platform.claude.com/docs/en/api/messages/batches/list). The API supports pagination, automatically fetching additional pages as needed:
608608 
609609<CodeGroup>
610610 ```bash cURL
from line 950
950950 
951951### Canceling a Message Batch
952952 
953You can cancel a Message Batch that is currently processing using the [cancel endpoint](https://platform.claude.com/docs/en/api/canceling-message-batches). Immediately after cancellation, a batch's `processing_status` will be `canceling`. You can use the same polling technique described earlier to wait until cancellation is finalized. Canceled batches end up with a status of `ended` and may contain partial results for requests that were processed before cancellation.
953You can cancel a Message Batch that is currently processing using the [cancel endpoint](https://platform.claude.com/docs/en/api/messages/batches/cancel). Immediately after cancellation, a batch's `processing_status` will be `canceling`. You can use the same polling technique described earlier to wait until cancellation is finalized. Canceled batches end up with a status of `ended` and may contain partial results for requests that were processed before cancellation.
954954 
955955<CodeGroup>
956956 ```bash cURL

build-with-claude/claude-on-amazon-bedrock-legacy Changed · +10 / -12 lines

from line 144
144144 
145145To invoke these models, pass an inference profile instead of the base model ID. The inference profile ID is the base model ID with a prefix from a column marked "Yes" in the following table, for example us.anthropic.claude-sonnet-4-5-20250929-v1:0. You can also pass the full inference profile ARN, in the form `arn:aws:bedrock:{region}:{account-id}:inference-profile/{inference-profile-id}`. For AWS's authoritative list of available inference profiles, see [Supported Regions and models for inference profiles](https://docs.aws.amazon.com/bedrock/latest/userguide/inference-profiles-support.html). To learn how the prefixes affect routing and pricing, see the [Global versus regional endpoints](https://platform.claude.com/docs/en/build-with-claude/claude-on-amazon-bedrock-legacy#global-vs-regional-endpoints) section.
146146 
147| Model | Base Bedrock model ID | `global` | `us` | `eu` | `jp` | `apac` |
148| ---------------------------- | ----------------------------------------- | -------- | ---- | ---- | ---- | ------ |
149| Claude Opus 4.6 | anthropic.claude-opus-4-6-v1 | Yes | Yes | Yes | Yes | Yes |
150| Claude Sonnet 4.6 | anthropic.claude-sonnet-4-6 | Yes | Yes | Yes | Yes | No |
151| Claude Sonnet 4.5 | anthropic.claude-sonnet-4-5-20250929-v1:0 | Yes | Yes | Yes | Yes | No |
152| Claude Sonnet 4 Deprecated. | anthropic.claude-sonnet-4-20250514-v1:0 | Yes | Yes | Yes | No | Yes |
153| Claude Sonnet 3.7 Retired. | anthropic.claude-3-7-sonnet-20250219-v1:0 | No | No | No | No | No |
154| Claude Opus 4.5 | anthropic.claude-opus-4-5-20251101-v1:0 | Yes | Yes | Yes | No | No |
155| Claude Opus 4.1 Deprecated. | anthropic.claude-opus-4-1-20250805-v1:0 | No | Yes | No | No | No |
156| Claude Opus 4 Retired. | anthropic.claude-opus-4-20250514-v1:0 | No | No | No | No | No |
157| Claude Haiku 4.5 | anthropic.claude-haiku-4-5-20251001-v1:0 | Yes | Yes | Yes | No | No |
158| Claude Haiku 3.5 Deprecated. | anthropic.claude-3-5-haiku-20241022-v1:0 | No | Yes | No | No | No |
147| Model | Base Bedrock model ID | `global` | `us` | `eu` | `jp` | `apac` |
148| :--------------------------------------------------------------------------------------------------- | :------------------------------------------ | :------- | :--- | :--- | :--- | :----- |
149| Claude Opus 4.6 | `anthropic.claude-opus-4-6-v1` | Yes | Yes | Yes | Yes | Yes |
150| Claude Opus 4.5 | `anthropic.claude-opus-4-5-20251101-v1:0` | Yes | Yes | Yes | No | No |
151| Claude Opus 4.1 ([deprecated](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | `anthropic.claude-opus-4-1-20250805-v1:0` | No | Yes | No | No | No |
152| Claude Sonnet 4.6 | `anthropic.claude-sonnet-4-6` | Yes | Yes | Yes | Yes | No |
153| Claude Sonnet 4.5 | `anthropic.claude-sonnet-4-5-20250929-v1:0` | Yes | Yes | Yes | Yes | No |
154| Claude Sonnet 4 ([deprecated](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | `anthropic.claude-sonnet-4-20250514-v1:0` | Yes | Yes | Yes | No | Yes |
155| Claude Haiku 4.5 | `anthropic.claude-haiku-4-5-20251001-v1:0` | Yes | Yes | Yes | No | No |
156| Claude Haiku 3.5 ([deprecated](https://platform.claude.com/docs/en/about-claude/model-deprecations)) | `anthropic.claude-3-5-haiku-20241022-v1:0` | No | Yes | No | No | No |
159157 
160158### List available models
161159 

build-with-claude/handling-stop-reasons Changed · +8 / -2 lines

from line 2394
23942394 
23952395 for _ in range(max_continuations):
23962396 response = client.messages.create(
2397 model="claude-opus-5-5", max_tokens=4096, messages=messages, tools=tools
2397 model="claude-opus-5-5",
2398 max_tokens=4096,
2399 messages=messages,
2400 tools=tools,
23982401 )
23992402 
24002403 if response.stop_reason != "pause_turn":
from line 3037
30343037 
30353038 while True:
30363039 response = client.messages.create(
3037 model="claude-opus-5-5", max_tokens=1024, messages=messages, tools=tools
3040 model="claude-opus-5-5",
3041 max_tokens=1024,
3042 messages=messages,
3043 tools=tools,
30383044 )
30393045 
30403046 if response.stop_reason == "tool_use":

build-with-claude/preserved-thinking Changed · +5 / -1 lines

from line 80
8080* **`"error"` (the default):** the API rejects the request with a 400 `invalid_request_error` that names the first failing block.
8181* **`"drop_block"`:** the API drops each failing block and every thinking block after it, and the request succeeds. Dropped blocks aren't billed. The model answers that turn without using reasoning from dropped blocks, and the prompt cache restarts at the edit. The response lists each dropped block in `input_transformations` (on the `message_start` event when streaming) with `reason: "prefix_binding_mismatch"`.
8282 
83`"drop_block"` keeps requests succeeding but doesn't fix the edit. Count the responses in each session whose `input_transformations` has a `prefix_binding_mismatch` entry, and alert on them. In the Message Batches API, an item that leaves the field unset doesn't fail. Where the API enforces the check by default, it drops the failing blocks instead. Set `"error"` explicitly there if you want batch items to fail.
83<Warning>
84 `"drop_block"` hides the error but doesn't fix the edit that caused it. Dropped blocks aren't billed, but a session's token usage might still increase because Claude can sometimes think more to re-create the dropped thinking. The increase tends to be larger when more thinking blocks are dropped, or when blocks are dropped on more turns of a long session.
85</Warning>
86 
87Count the responses in each session whose `input_transformations` has a `prefix_binding_mismatch` entry, alert on them, and replace each edit with the matching pattern in [Make changes without editing the prefix](https://platform.claude.com/docs/en/build-with-claude/preserved-thinking#replace-prefix-edits). In the Message Batches API, an item that leaves the field unset doesn't fail. Where the API enforces the check by default, it drops the failing blocks instead. Set `"error"` explicitly there if you want batch items to fail.
8488 
8589Both the field and the `input_transformations` array require the `thinking-binding-controls-2026-08-01` [beta header](https://platform.claude.com/docs/en/api/beta-headers). [Set the mismatch behavior and read `input_transformations`](https://platform.claude.com/docs/en/build-with-claude/preserved-thinking#preserved-thinking-controls) shows the request in each SDK.
8690 

build-with-claude/refusals-and-fallback Changed · +17 / -9 lines

from line 180
180180* `category` and `explanation` are both `null` when the refusal does not map to a named category. That `null` is a normal, permanent value, not a placeholder.
181181* `stop_details` itself is `null` for every stop reason other than `refusal`.
182182 
183| `category` | What it means |
184| ------------------------ | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
185| `"cyber"` | The request could enable cyber harm, such as malware or exploit development. Benign cybersecurity work can also trigger this category. |
186| `"bio"` | The request could enable biological harm, such as dangerous lab methods. Beneficial life sciences work can also trigger this category. |
187| `"frontier_llm"` | The request could assist the development of competing AI models, which is restricted under [Anthropic's commercial terms](https://www.anthropic.com/legal/commercial-terms). Benign machine learning work can also trigger this category. |
188| `"reasoning_extraction"` | The request asks the model to reproduce its internal reasoning in the response text. To get reasoning in a structured form instead, use [adaptive thinking](https://platform.claude.com/docs/en/build-with-claude/thinking). |
189| `"general_harms"` | The request falls under a usage-policy area outside the four named categories. Benign work can also trigger this category. |
183| `category` | What it means | Billed before any output |
184| ------------------------ | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------ |
185| `"cyber"` | The request could enable cyber harm, such as malware or exploit development. Benign cybersecurity work can also trigger this category. | No |
186| `"bio"` | The request could enable biological harm, such as dangerous lab methods. Beneficial life sciences work can also trigger this category. | Yes |
187| `"frontier_llm"` | The request could assist the development of competing AI models, which is restricted under [Anthropic's commercial terms](https://www.anthropic.com/legal/commercial-terms). Benign machine learning work can also trigger this category. | Yes |
188| `"reasoning_extraction"` | The request asks the model to reproduce its internal reasoning in the response text. To get reasoning in a structured form instead, use [adaptive thinking](https://platform.claude.com/docs/en/build-with-claude/thinking). | Yes |
189| `"general_harms"` | The request falls under a usage-policy area outside the four named categories. Benign work can also trigger this category. | No |
190190 
191191A refusal can arrive before any output, or mid-stream after partial output. In either case, treat any partial output as incomplete and discard it.
192192 
193193## How refusals are billed
194194 
195You are not billed for a refusal that arrives before any output. `content` is empty, and token counts appear in `usage` but are not charged. The request still counts against your rate limits. A mid-stream refusal bills the input tokens and the output already streamed at normal rates.
195These billing rules apply on every platform: the Claude API, Amazon Bedrock, Claude Platform on AWS, Google Cloud, and Microsoft Foundry.
196196 
197**Refusals before any output:** To disrupt attempts to circumvent Anthropic's safeguards at scale, a refusal that arrives before any output is billed when its `stop_details.category` is `"bio"`, `"frontier_llm"`, or `"reasoning_extraction"`. These are the categories where Anthropic measures low volumes of false positives, as of September 2026. These refusals are billed like any other request, at the rates of the model that ran it. A refusal before any output in any other category, or with a `null` category, is not billed. Either way, `content` is empty and token counts appear in `usage`. The request still counts against your rate limits.
198 
199**Mid-stream refusals:** A mid-stream refusal bills the input tokens and the output already streamed at normal rates.
200 
201**Fallback:** When you use fallback, the refusal that triggered it is billed, in addition to the fallback request, when it arrived mid-stream or is in one of the billed categories. [Fallback credit](https://platform.claude.com/docs/en/build-with-claude/fallback-credit) compensates for the fallback request's prompt-cache miss, so you don't pay to cache the conversation twice. For how server-side fallback reports each attempt, see [Billing and rate limits](https://platform.claude.com/docs/en/build-with-claude/refusals-and-fallback#billing-and-rate-limits).
202 
203The billed categories may change as Anthropic keeps measuring and refining its safeguards' false positive rates. The **Billed before any output** column in the [refusal category table](https://platform.claude.com/docs/en/build-with-claude/refusals-and-fallback#refusal-response) lists the billed categories.
204 
197205## Picking a fallback approach
198206 
199207There are three ways to retry a refused request on another model. The right one depends on where you are running and how much control you need.
from line 745
737745 
738746### Billing and rate limits
739747 
740An attempt that declined before producing any output is not billed: its tokens are reported on its `usage.iterations` entry but not charged. Every attempt that produced output, including one that declined partway through its response, is billed separately at the rates of the model that ran it. The `usage.iterations` array is the per-attempt record of what you're billed. The top-level `usage` counts describe only the attempt that produced the returned message. Tokens from different models are never summed into one field.
748Each attempt follows the rules in [How refusals are billed](https://platform.claude.com/docs/en/build-with-claude/refusals-and-fallback#how-refusals-are-billed), at the rates of the model that ran it. An attempt that declined before producing any output is billed only when its refusal category is billed, and its tokens are reported on its `usage.iterations` entry either way. Every attempt that produced output, including one that declined partway through its response, is billed separately. The `usage.iterations` array is the per-attempt record of what you're billed. The top-level `usage` counts describe only the attempt that produced the returned message. Tokens from different models are never summed into one field.
741749 
742750Every attempt that runs, including one that declined, counts against its own model's rate limits.
743751 

cli-sdks-libraries/cli/apply Changed · +22 / -20 lines

from line 12
1212 
1313Write the agent as a Markdown file under `agents/` and apply it:
1414 
15<MultiFileExample language="cli" label="CLI">
16 ```bash CLI
17 ant apply agents/summarizer.md
18 ```
15<CodeGroup>
16 <CodeGroupItem>
17 ```bash CLI
18 ant apply agents/summarizer.md
19 ```
1920 
20 <File filename="agents/summarizer.md">
21 ```markdown
22 ---
23 name: Summarizer
24 model: claude-opus-5-5
25 tools:
26 - type: agent_toolset_20260401
27 ---
21 <File filename="agents/summarizer.md">
22 ```markdown
23 ---
24 name: Summarizer
25 model: claude-opus-5-5
26 tools:
27 - type: agent_toolset_20260401
28 ---
2829 
29 You are a helpful assistant that writes concise summaries.
30 ```
31 </File>
32</MultiFileExample>
30 You are a helpful assistant that writes concise summaries.
31 ```
32 </File>
33 </CodeGroupItem>
34</CodeGroup>
3335 
3436The frontmatter holds the agent's configuration (the fields from [Define your agent](https://platform.claude.com/docs/en/managed-agents/agent-setup)) and the body is its system prompt. `ant apply` [infers](https://platform.claude.com/docs/en/cli-sdks-libraries/cli/apply#kind-inference) that the file is an agent from its path, here the `agents/` directory.
3537 
from line 108
106108 
107109Resources refer to each other by path. Wherever the API expects another resource's ID, write the relative path to that resource's file instead. In this project, the reviewer agent lists `../skills/pr-summary` under `skills`, the lead agent lists `./reviewer.md` in its roster, and the deployment names its agent, environment, and memory store by path. `ant apply` creates them in dependency order and fills in the real IDs. The project has six files:
108110 
109<MultiFileExample variant="explorer">
111<FileExplorer>
110112 <File filename="agents/reviewer.md">
111113 ```markdown
112114 ---
from line 190
188190 Review any open pull requests. Start with the oldest.
189191 ```
190192 </File>
191</MultiFileExample>
193</FileExplorer>
192194 
193195Apply the whole directory:
194196 
Feedback