Sweep 22 Sep 2026 · 17:19Z Build v2.1.280 501 read Stable v2.1.267 Latest v2.1.280 Next v2.1.280 Feeds RSS JSON llms.txt Unofficial
One change · api

Update Agent changed

api/beta/agents/update

Nearest release: v2.1.268, published under an hour after this site recorded the change. Shown because the two are within 24 hours of each other. Nothing here says the release caused the edit.

Recorded here
Lines+1,000added
Lines−899removed
From line 1 where the diff opens
First seen 14 Aug 2026 this site's first read of the page
Recorded edits9to this page, all time

The whole hunk

from line 1, old and new numbered
/
lines
from line 1
1---
2title: Update Agent
3url: https://platform.claude.com/docs/en/api/beta/agents/update
4---
5 
16# Update Agent
27 
38**POST** `/v1/agents/{agent_id}`
from line 21
1621 
1722 - `string`
1823 
19 - `"message-batches-2024-09-24" or "prompt-caching-2024-07-31" or "computer-use-2024-10-22" or 41 more`
24 - `"message-batches-2024-09-24" or "prompt-caching-2024-07-31" or "computer-use-2024-10-22" or 42 more`
2025 
2126 - `"message-batches-2024-09-24"`
2227 
from line 69
6469 
6570 - `"user-profiles-2026-08-18"`
6671 
72 - `"user-profiles-2026-09-04"`
73 
6774 - `"advisor-tool-2026-03-01"`
6875 
6976 - `"managed-agents-2026-04-01"`
from line 113
106113 
107114 - `"mid-conversation-system-clear-at-2026-08-21"`
108115 
116- `"anthropic-workspace-id": optional string`
117 
109118## Body parameters
110119 
111120- `description: optional string or null`
from line 127
118127 
119128 MCP servers. Full replacement. Omit to preserve; send empty array or `null` to clear. Names must be unique. Maximum 20. Every server must be referenced by an `mcp_toolset` in the agent's resulting `tools`; unreferenced servers are rejected. See the [MCP connector guide](https://platform.claude.com/docs/en/managed-agents/mcp-connector).
120129 
130 - `type: "url"`
131 
121132 - `name: string`
122133 
123134 Unique name for this server, referenced by mcp_toolset configurations. 1-255 characters.
from line 135
124135 
125136 minLength: 1, maxLength: 255
126137 
127 - `type: "url"`
128 
129138 - `url: string`
130139 
131140 Endpoint URL for the MCP server.
from line 293
284293 
285294 A coordinator topology: the session's primary thread orchestrates work by spawning session threads, each running an agent drawn from the `agents` roster.
286295 
296 - `type: "coordinator"`
297 
287298 - `agents: array of BetaManagedAgentsMultiagentRosterEntryParams`
288299 
289300 Agents the coordinator may spawn as session threads. 1–20 entries. Each entry is an agent ID string, a versioned `{"type":"agent","id","version"}` reference, or `{"type":"self"}` to allow recursive self-invocation. Entries must reference distinct agents (after resolving `self` and string forms); at most one `self`. Referenced agents must exist, must not be archived, and must not themselves have `multiagent` set (depth limit 1).
from line 305
294305 
295306 Specification for an Agent. Provide a specific `version` or use the short-form `agent="agent_id"` for the most recent version
296307 
308 - `type: "agent"`
309 
297310 - `id: string`
298311 
299312 The `agent` ID.
from line 313
300313 
301314 minLength: 1, maxLength: 128
302315 
303 - `type: "agent"`
304 
305316 - `version: optional number`
306317 
307318 The specific `agent` version to use. Omit to use the latest version. Must be at least 1 if specified.
from line 329
318329 
319330 Platform advisor roster entry: a model the session's primary thread may consult mid-turn. At most one per roster; the entry occupies the roster name `anthropic.advisor`.
320331 
332 - `type: "advisor"`
333 
321334 - `model: string`
322335 
323336 A Claude model id. The model must be permitted as an advisor for this agent's model — see the sessions/threads/advisor spec.
from line 337
324337 
325338 minLength: 1, maxLength: 256
326339 
327 - `type: "advisor"`
328 
329 - `type: "coordinator"`
330 
331340- `name: optional string`
332341 
333342 Human-readable name. Must be non-empty. Omit to preserve. Cannot be cleared.
from line 351
342351 
343352 An Anthropic-managed skill.
344353 
354 - `type: "anthropic"`
355 
345356 - `skill_id: string`
346357 
347358 Identifier of the Anthropic skill (e.g., "xlsx").
from line 359
348359 
349360 minLength: 1, maxLength: 64
350361 
351 - `type: "anthropic"`
352 
353362 - `version: optional string or null`
354363 
355364 Version to pin. Defaults to latest if omitted.
from line 369
360369 
361370 A user-created custom skill.
362371 
372 - `type: "custom"`
373 
363374 - `skill_id: string`
364375 
365376 Tagged ID of the custom skill (e.g., "skill_01XJ5...").
from line 377
366377 
367378 minLength: 1, maxLength: 64
368379 
369 - `type: "custom"`
370 
371380 - `version: optional string or null`
372381 
373382 Version to pin. Defaults to latest if omitted.
from line 407
398407 
399408 Configuration override for the bash tool.
400409 
410 - `type: optional "bash"`
411 
401412 - `name: "bash"`
402413 
403414 Must be "bash".
from line 417
406417 
407418 Whether this tool is enabled and available to Claude. Overrides the default_config setting.
408419 
409 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`
420 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy or null`
410421 
411422 Permission policy for tool execution.
412423 
from line 433
422433 
423434 - `type: "always_ask"`
424435 
425 - `type: optional "bash"`
436 - `BetaManagedAgentsAutoPolicy object`
426437 
438 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
439 
440 - `type: "auto"`
441 
427442 - `BetaManagedAgentsEditToolConfigParams object`
428443 
429444 Configuration override for the edit tool.
430445 
446 - `type: optional "edit"`
447 
431448 - `name: "edit"`
432449 
433450 Must be "edit".
from line 453
436453 
437454 Whether this tool is enabled and available to Claude. Overrides the default_config setting.
438455 
439 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`
456 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy or null`
440457 
441458 Permission policy for tool execution.
442459 
from line 465
448465 
449466 Tool calls require user confirmation before execution.
450467 
451 - `type: optional "edit"`
468 - `BetaManagedAgentsAutoPolicy object`
452469 
470 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
471 
453472 - `BetaManagedAgentsReadToolConfigParams object`
454473 
455474 Configuration override for the read tool.
456475 
476 - `type: optional "read"`
477 
457478 - `name: "read"`
458479 
459480 Must be "read".
from line 483
462483 
463484 Whether this tool is enabled and available to Claude. Overrides the default_config setting.
464485 
465 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`
486 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy or null`
466487 
467488 Permission policy for tool execution.
468489 
from line 495
474495 
475496 Tool calls require user confirmation before execution.
476497 
477 - `type: optional "read"`
498 - `BetaManagedAgentsAutoPolicy object`
478499 
500 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
501 
479502 - `BetaManagedAgentsWriteToolConfigParams object`
480503 
481504 Configuration override for the write tool.
482505 
506 - `type: optional "write"`
507 
483508 - `name: "write"`
484509 
485510 Must be "write".
from line 513
488513 
489514 Whether this tool is enabled and available to Claude. Overrides the default_config setting.
490515 
491 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`
516 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy or null`
492517 
493518 Permission policy for tool execution.
494519 
from line 525
500525 
501526 Tool calls require user confirmation before execution.
502527 
503 - `type: optional "write"`
528 - `BetaManagedAgentsAutoPolicy object`
504529 
530 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
531 
505532 - `BetaManagedAgentsGlobToolConfigParams object`
506533 
507534 Configuration override for the glob tool.
508535 
536 - `type: optional "glob"`
537 
509538 - `name: "glob"`
510539 
511540 Must be "glob".
from line 543
514543 
515544 Whether this tool is enabled and available to Claude. Overrides the default_config setting.
516545 
517 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`
546 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy or null`
518547 
519548 Permission policy for tool execution.
520549 
from line 555
526555 
527556 Tool calls require user confirmation before execution.
528557 
529 - `type: optional "glob"`
558 - `BetaManagedAgentsAutoPolicy object`
530559 
560 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
561 
531562 - `BetaManagedAgentsGrepToolConfigParams object`
532563 
533564 Configuration override for the grep tool.
534565 
566 - `type: optional "grep"`
567 
535568 - `name: "grep"`
536569 
537570 Must be "grep".
from line 573
540573 
541574 Whether this tool is enabled and available to Claude. Overrides the default_config setting.
542575 
543 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`
576 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy or null`
544577 
545578 Permission policy for tool execution.
546579 
from line 585
552585 
553586 Tool calls require user confirmation before execution.
554587 
555 - `type: optional "grep"`
588 - `BetaManagedAgentsAutoPolicy object`
556589 
590 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
591 
557592 - `BetaManagedAgentsWebFetchToolConfigParams object`
558593 
559594 Configuration override for the web_fetch tool.
560595 
596 - `type: optional "web_fetch"`
597 
561598 - `name: "web_fetch"`
562599 
563600 Must be "web_fetch".
from line 617
580617 
581618 format: int32
582619 
583 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`
620 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy or null`
584621 
585622 Permission policy for tool execution.
586623 
from line 629
592629 
593630 Tool calls require user confirmation before execution.
594631 
595 - `type: optional "web_fetch"`
632 - `BetaManagedAgentsAutoPolicy object`
596633 
634 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
635 
597636 - `BetaManagedAgentsWebSearchToolConfigParams object`
598637 
599638 Configuration override for the web_search tool.
600639 
640 - `type: optional "web_search"`
641 
601642 - `name: "web_search"`
602643 
603644 Must be "web_search".
from line 655
614655 
615656 Whether this tool is enabled and available to Claude. Overrides the default_config setting.
616657 
617 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`
658 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy or null`
618659 
619660 Permission policy for tool execution.
620661 
from line 667
626667 
627668 Tool calls require user confirmation before execution.
628669 
629 - `type: optional "web_search"`
670 - `BetaManagedAgentsAutoPolicy object`
630671 
672 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
673 
631674 - `user_location: optional BetaManagedAgentsUserLocation or null`
632675 
633676 Approximate user location for search result localization.
from line 709
666709 
667710 Whether tools are enabled and available to Claude by default. Defaults to true if not specified.
668711 
669 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`
712 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy or null`
670713 
671714 Permission policy for tool execution.
672715 
from line 721
678721 
679722 Tool calls require user confirmation before execution.
680723 
724 - `BetaManagedAgentsAutoPolicy object`
725 
726 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
727 
681728 - `BetaManagedAgentsMCPToolsetParams object`
682729 
683730 Configuration for tools from an MCP server defined in `mcp_servers`.
684731 
732 - `type: "mcp_toolset"`
733 
685734 - `mcp_server_name: string`
686735 
687736 Name of the MCP server. Must match a server name from the mcp_servers array. 1-255 characters.
from line 737
688737 
689738 minLength: 1, maxLength: 255
690739 
691 - `type: "mcp_toolset"`
692 
693740 - `configs: optional array of BetaManagedAgentsMCPToolConfigParams`
694741 
695742 Per-tool configuration overrides.
from line 751
704751 
705752 Whether this tool is enabled. Overrides the `default_config` setting.
706753 
707 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`
754 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy or null`
708755 
709756 Permission policy for tool execution.
710757 
from line 763
716763 
717764 Tool calls require user confirmation before execution.
718765 
766 - `BetaManagedAgentsAutoPolicy object`
767 
768 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
769 
719770 - `default_config: optional BetaManagedAgentsMCPToolsetDefaultConfigParams or null`
720771 
721772 Default configuration for all tools from an MCP server.
from line 775
724775 
725776 Whether tools are enabled by default. Defaults to true if not specified.
726777 
727 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`
778 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy or null`
728779 
729780 Permission policy for tool execution.
730781 
from line 787
736787 
737788 Tool calls require user confirmation before execution.
738789 
790 - `BetaManagedAgentsAutoPolicy object`
791 
792 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
793 
739794 - `BetaManagedAgentsCustomToolParams object`
740795 
741796 A custom tool that is executed by the API client rather than the agent. When the agent calls this tool, an `agent.custom_tool_use` event is emitted and the session goes idle, waiting for the client to provide the result via a `user.custom_tool_result` event.
742797 
798 - `type: "custom"`
799 
743800 - `description: string`
744801 
745802 Description of what the tool does, shown to the agent to help it decide when to use the tool.
from line 819
762819 
763820 minLength: 1, maxLength: 128
764821 
765 - `type: "custom"`
766 
767822- `version: optional number`
768823 
769824 The agent's current version, used to prevent concurrent overwrites. Obtain this value from a create or retrieve response. Must be at least 1 if specified. When supplied, the request fails if it does not match the server's current version; omit to apply the update unconditionally.
from line 831
776831 
777832 A Managed Agents `agent`.
778833 
834 - `type: "agent"`
835 
779836 - `id: string`
780837 
781838 - `archived_at: string or null`
from line 851
794851 
795852 - `mcp_servers: array of BetaManagedAgentsMCPServerURLDefinition`
796853 
797 - `name: string`
798 
799854 - `type: "url"`
800855 
856 - `name: string`
857 
801858 - `url: string`
802859 
803860 - `metadata: map[string]`
from line 983
926983 
927984 Resolved coordinator topology with a concrete agent roster.
928985 
986 - `type: "coordinator"`
987 
929988 - `agents: array of BetaManagedAgentsAgentReference or BetaManagedAgentsAdvisor`
930989 
931990 Agents the coordinator may spawn as session threads, each resolved to a specific version.
from line 993
934993 
935994 A resolved agent reference with a concrete version.
936995 
937 - `id: string`
938 
939996 - `type: "agent"`
940997 
998 - `id: string`
999 
9411000 - `version: number`
9421001 
9431002 format: int32
from line 1005
9461005 
9471006 Platform advisor roster entry: a model the session's primary thread may consult mid-turn.
9481007 
1008 - `type: "advisor"`
1009 
9491010 - `model: string`
9501011 
9511012 The advisor model id.
9521013 
953 - `type: "advisor"`
954 
955 - `type: "coordinator"`
956 
9571014 - `name: string`
9581015 
9591016 - `skills: array of BetaManagedAgentsAnthropicSkill or BetaManagedAgentsCustomSkill`
from line 1019
9621019 
9631020 A resolved Anthropic-managed skill.
9641021 
965 - `skill_id: string`
966 
9671022 - `type: "anthropic"`
9681023 
1024 - `skill_id: string`
1025 
9691026 - `version: string`
9701027 
9711028 - `BetaManagedAgentsCustomSkill object`
from line 1029
9721029 
9731030 A resolved user-created custom skill.
9741031 
975 - `skill_id: string`
976 
9771032 - `type: "custom"`
9781033 
1034 - `skill_id: string`
1035 
9791036 - `version: string`
9801037 
9811038 - `system: string or null`
from line 1041
9841041 
9851042 - `BetaManagedAgentsAgentToolset20260401 object`
9861043 
1044 - `type: "agent_toolset_20260401"`
1045 
9871046 - `configs: array of BetaManagedAgentsAgentToolConfig`
9881047 
9891048 - `BetaManagedAgentsBashToolConfig object`
from line 1049
9901049 
9911050 Configuration for the bash tool.
9921051 
1052 - `type: "bash"`
1053 
9931054 - `enabled: boolean`
9941055 
9951056 - `name: "bash"`
9961057 
997 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`
1058 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`
9981059 
9991060 Permission policy for tool execution.
10001061 
from line 1071
10101071 
10111072 - `type: "always_ask"`
10121073 
1013 - `type: "bash"`
1074 - `BetaManagedAgentsAutoPolicy object`
10141075 
1076 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
1077 
1078 - `type: "auto"`
1079 
10151080 - `BetaManagedAgentsEditToolConfig object`
10161081 
10171082 Configuration for the edit tool.
10181083 
1084 - `type: "edit"`
1085 
10191086 - `enabled: boolean`
10201087 
10211088 - `name: "edit"`
10221089 
1023 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`
1090 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`
10241091 
10251092 Permission policy for tool execution.
10261093 
from line 1099
10321099 
10331100 Tool calls require user confirmation before execution.
10341101 
1035 - `type: "edit"`
1102 - `BetaManagedAgentsAutoPolicy object`
10361103 
1104 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
1105 
10371106 - `BetaManagedAgentsReadToolConfig object`
10381107 
10391108 Configuration for the read tool.
10401109 
1110 - `type: "read"`
1111 
10411112 - `enabled: boolean`
10421113 
10431114 - `name: "read"`
10441115 
1045 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`
1116 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`
10461117 
10471118 Permission policy for tool execution.
10481119 
from line 1125
10541125 
10551126 Tool calls require user confirmation before execution.
10561127 
1057 - `type: "read"`
1128 - `BetaManagedAgentsAutoPolicy object`
10581129 
1130 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
1131 
10591132 - `BetaManagedAgentsWriteToolConfig object`
10601133 
10611134 Configuration for the write tool.
10621135 
1136 - `type: "write"`
1137 
10631138 - `enabled: boolean`
10641139 
10651140 - `name: "write"`
10661141 
1067 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`
1142 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`
10681143 
10691144 Permission policy for tool execution.
10701145 
from line 1151
10761151 
10771152 Tool calls require user confirmation before execution.
10781153 
1079 - `type: "write"`
1154 - `BetaManagedAgentsAutoPolicy object`
10801155 
1156 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
1157 
10811158 - `BetaManagedAgentsGlobToolConfig object`
10821159 
10831160 Configuration for the glob tool.
10841161 
1162 - `type: "glob"`
1163 
10851164 - `enabled: boolean`
10861165 
10871166 - `name: "glob"`
10881167 
1089 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`
1168 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`
10901169 
10911170 Permission policy for tool execution.
10921171 
from line 1177
10981177 
10991178 Tool calls require user confirmation before execution.
11001179 
1101 - `type: "glob"`
1180 - `BetaManagedAgentsAutoPolicy object`
11021181 
1182 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
1183 
11031184 - `BetaManagedAgentsGrepToolConfig object`
11041185 
11051186 Configuration for the grep tool.
11061187 
1188 - `type: "grep"`
1189 
11071190 - `enabled: boolean`
11081191 
11091192 - `name: "grep"`
11101193 
1111 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`
1194 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`
11121195 
11131196 Permission policy for tool execution.
11141197 
from line 1203
11201203 
11211204 Tool calls require user confirmation before execution.
11221205 
1123 - `type: "grep"`
1206 - `BetaManagedAgentsAutoPolicy object`
11241207 
1208 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
1209 
11251210 - `BetaManagedAgentsWebFetchToolConfig object`
11261211 
11271212 Configuration for the web_fetch tool.
11281213 
1214 - `type: "web_fetch"`
1215 
11291216 - `enabled: boolean`
11301217 
11311218 - `name: "web_fetch"`
11321219 
1133 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`
1220 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`
11341221 
11351222 Permission policy for tool execution.
11361223 
from line 1229
11421229 
11431230 Tool calls require user confirmation before execution.
11441231 
1145 - `type: "web_fetch"`
1232 - `BetaManagedAgentsAutoPolicy object`
11461233 
1234 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
1235 
11471236 - `allowed_domains: optional array of string`
11481237 
11491238 - `blocked_domains: optional array of string`
from line 1245
11561245 
11571246 Configuration for the web_search tool.
11581247 
1248 - `type: "web_search"`
1249 
11591250 - `enabled: boolean`
11601251 
11611252 - `name: "web_search"`
11621253 
1163 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`
1254 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`
11641255 
11651256 Permission policy for tool execution.
11661257 
from line 1263
11721263 
11731264 Tool calls require user confirmation before execution.
11741265 
1175 - `type: "web_search"`
1266 - `BetaManagedAgentsAutoPolicy object`
11761267 
1268 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
1269 
11771270 - `allowed_domains: optional array of string`
11781271 
11791272 - `blocked_domains: optional array of string`
from line 1307
12141307 
12151308 - `enabled: boolean`
12161309 
1217 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`
1310 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`
12181311 
12191312 Permission policy for tool execution.
12201313 
from line 1319
12261319 
12271320 Tool calls require user confirmation before execution.
12281321 
1229 - `type: "agent_toolset_20260401"`
1322 - `BetaManagedAgentsAutoPolicy object`
12301323 
1324 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
1325 
12311326 - `BetaManagedAgentsMCPToolset object`
12321327 
1328 - `type: "mcp_toolset"`
1329 
12331330 - `configs: array of BetaManagedAgentsMCPToolConfig`
12341331 
12351332 - `enabled: boolean`
from line 1333
12361333 
12371334 - `name: string`
12381335 
1239 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`
1336 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`
12401337 
12411338 Permission policy for tool execution.
12421339 
from line 1345
12481345 
12491346 Tool calls require user confirmation before execution.
12501347 
1348 - `BetaManagedAgentsAutoPolicy object`
1349 
1350 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
1351 
12511352 - `default_config: BetaManagedAgentsMCPToolsetDefaultConfig`
12521353 
12531354 Resolved default configuration for all tools from an MCP server.
from line 1355
12541355 
12551356 - `enabled: boolean`
12561357 
1257 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`
1358 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or BetaManagedAgentsAutoPolicy`
12581359 
12591360 Permission policy for tool execution.
12601361 
from line 1367
12661367 
12671368 Tool calls require user confirmation before execution.
12681369 
1269 - `mcp_server_name: string`
1370 - `BetaManagedAgentsAutoPolicy object`
12701371 
1271 - `type: "mcp_toolset"`
1372 The server decides each tool call individually: it judges, from the tool, its input, and the session content so far, whether the call is safe to execute or high-risk, and evaluates it to allow when judged safe and to deny when judged high-risk. A call the server cannot reach a judgement on evaluates to ask.
12721373 
1374 - `mcp_server_name: string`
1375 
12731376 - `BetaManagedAgentsCustomTool object`
12741377 
12751378 A custom tool as returned in API responses.
12761379 
1380 - `type: "custom"`
1381 
12771382 - `description: string`
12781383 
12791384 - `input_schema: BetaManagedAgentsCustomToolInputSchema`
from line 1392
12871392 - `required: optional array of string or null`
12881393 
12891394 - `name: string`
1290 
1291 - `type: "custom"`
1292 
1293 - `type: "agent"`
12941395 
12951396 - `updated_at: string`
12961397