One read of Claude Developer Platformapi-20260924T173729Z
83 pages moved out of 637 read.
Pages moved
83
significant first
Pages read
637
in this capture
Captured
17:37 UTC
Corpus hash
60090eb3549b
corpus-hash
What this read moved
1-25 of 83, page 1 of 4This capture is too large to show at once. Changes 1-25 of 83 are below, significant first; the rest are on the following screens.
about-claude/models/optimizing-for-cost-and-intelligence Changed · +103 / -2 lines
### Show the model elapsed time
from line 13
1313The levers come in two kinds:
1414
1515* **Free wins** cut spend without touching quality: prompt caching, token hygiene, a prompt audit against the model you are running, [batch processing](https://platform.claude.com/docs/en/build-with-claude/batch-processing) at 50% off for work that can wait up to 24 hours, and [workspace spend limits](https://platform.claude.com/docs/en/api/rate-limits#setting-lower-limits-for-workspaces) as the backstop.
16* **Tradeoffs** exchange cost for intelligence: model choice, effort, output caps and task budgets, and multi-model architectures.
16* **Tradeoffs** exchange cost for intelligence: model choice, effort, output caps and task budgets, an elapsed-time clock, and multi-model architectures.
1717
1818Each lever comes with measured results and the rule for when it pays. In Anthropic's measurements, prompt caching was the largest lever by a wide margin: it cut agent-loop cost by a factor of 2.7 to 5.3 on this guide's benchmarks and cut a small triage agent's bill by 83%, or 88% with input trimming added. The multi-model levers are narrower; a second model paid off in two shapes, an advisor and an orchestrator.
1919
from line 32
3232| Attempts end with `stop_reason: max_tokens` | Raise `max_tokens`; 64,000 covered all but 2 of 14,000 turns measured at the default effort, and 128,000 cost nothing extra per solved task | [Set budgets](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#set-budgets-and-output-caps) |
3333| You can check outputs (tests, a verifier) | Run everything at low effort and re-run failures at `high`; on the coding benchmark measured, the pass rate held at about half the cost | [Re-run failures](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#re-run-failures-at-higher-effort) |
3434| Agent loops with a few very costly runs | Set a task budget (beta; check the support table for which models), a Claude Managed Agents session budget, and a workspace spend limit | [Set budgets](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#set-budgets-and-output-caps) |
35| You want agent runs to finish sooner | Tell the model that time matters, and show it the elapsed time; on DRACO, HLE, and an internal physics set, runs took 33% to 69% less time at a 28% to 54% lower cost per task, with scores up to 1.9 points lower | [Show the model elapsed time](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#show-the-model-elapsed-time) |
3536| A lower-cost model stalls only on hard decisions | Add a frontier advisor. It pays off when priced well above the executor and actually consulted, so first price the advisor's model alone at low effort and measure the consult rate | [Advisor strategy](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#advisor-strategy-escalate-hard-decisions) |
3637| The work exceeds one context window | Delegate partitions to cheaper workers | [Orchestrator strategy](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#orchestrator-strategy-delegate-bulk-work) |
3738
from line 244
243244
244245## Trade cost against intelligence
245246
246These levers set where a single model sits between cost and intelligence: model choice, effort, re-running failures at a higher setting, and the budgets and caps it works within. Start with an effort sweep on your current model ([Tune effort](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#tune-effort)). From lowest to highest cost and capability, the current models are Claude Haiku 4.5, Claude Sonnet 5, Claude Opus 5.5, and Claude Fable 5.1 (the frontier model); [Models overview](https://platform.claude.com/docs/en/models/overview) has the full lineup and prices.
247These levers set where a single model sits between cost and intelligence: model choice, effort, re-running failures at a higher setting, the budgets and caps it works within, and whether it can see how much time has passed. Start with an effort sweep on your current model ([Tune effort](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#tune-effort)). From lowest to highest cost and capability, the current models are Claude Haiku 4.5, Claude Sonnet 5, Claude Opus 5.5, and Claude Fable 5.1 (the frontier model); [Models overview](https://platform.claude.com/docs/en/models/overview) has the full lineup and prices.
247248
248249### Compare models on cost per task
249250
from line 363
362363
363364
364365
366### Show the model elapsed time
367
368A model in an agent loop can't see a clock. A [task budget](https://platform.claude.com/docs/en/build-with-claude/task-budgets) shows it how many tokens are left, but by default nothing in the request shows it how long the work has taken. Two small changes give it that signal. Add a two-sentence instruction to the system prompt that says time matters, and from the second request on, send the elapsed time before each of the model's turns.
369
370Anthropic measured both changes together with Claude Fable 5.1 at `high` effort, on two public benchmarks, DRACO[21](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#refs) and HLE[22](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#refs), and on an internal set of 70 research-level physics problems, adapted from the public CritPt benchmark[23](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#refs). This page calls that set the physics set. Each of the three ran in two shapes: a single agent, and a team in which a lead agent starts helper agents of the same model that work in parallel. A score change counts as inside the margin when its 95% interval stays within a limit that Anthropic set before the runs: 1.5 points on DRACO and 2.5 points on HLE. The following chart plots score against cost per task for each configuration. A second row of bars gives each configuration's time as a ratio to the single agent at `high` effort, without retry waits. A third row gives the score change that both changes make, with its 95% interval:
371
372
373
374**With a team of agents.** A team does more work than a single agent, so by default it costs more. On DRACO, the team cost 4.0 times as much as the single agent and took about as long (95% interval 12% less to 13% more). With the instruction and the clock on every agent, the team finished in 33% less time at a 54% lower cost per task. Its score was 1.5 points lower (95% interval 0.9 to 2.1 lower), and the far end of that interval, 2.1 points lower, is past the 1.5-point margin. On HLE, the team finished in 51% less time at a 54% lower cost per task. Its score was 1.7 points lower (95% interval 0.3 to 3.1 lower), and the far end of that interval, 3.1 points lower, is past the 2.5-point margin. On the physics set[23](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#refs), the team finished in 39% less time. Its cost per task was 28% lower, and that saving depends on how often the prompt cache expired between requests. With no expiry, it would be 23%. Its score was 0.2 points higher (95% interval 1.5 lower to 2.0 higher).
375
376On DRACO, the lead started a median of 4 helpers per attempt, so the DRACO result shows a team working in parallel. On HLE and the physics set, the lead started a median of 0 helpers, so at least half of those team runs had only the lead agent. Those team results mostly show the lead agent's own behavior, not the effect of parallel helpers.
377
378**With a single agent.** On the physics set[23](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#refs), the same changes cut a single agent's time by 34% and its cost per task by 34%. Its score was 0.2 points lower (95% interval 2.5 lower to 2.1 higher). On the physics set, a lower effort level saved cost but not clearly time. At `medium` effort, the single agent cost 37% less per task than at `high`, and its time was 9% less (95% interval 30% less to 16% more). It scored 3.4 points lower (95% interval 0.4 to 6.8 lower), and the interval reaches close to zero. With both changes at `high` effort, the single agent took 27% less time than at `medium` effort (95% interval 5% less to 44% less). Its cost per task was 6% more (95% interval 12% less to 27% more), and its score was 3.2 points higher (95% interval 0.1 lower to 6.5 higher).
379
380On HLE, the same changes cut a single agent's time by 54% and its cost per task by 48%. Its score was 1.1 points lower (95% interval 2.6 lower to 0.3 higher), and the far end of that interval, 2.6 points lower, is just past the 2.5-point margin. At `medium` effort, the single agent cost 43% less per task than at `high`, took 39% less time, and scored 1.3 points lower (95% interval 2.8 lower to 0.1 higher). With both changes at `high` effort, the single agent took 25% less time than at `medium` effort (95% interval 12% less to 35% less). Its cost per task was 9% less (95% interval 21% less to 6% more), and its score was 0.2 points higher (95% interval 1.3 lower to 1.7 higher).
381
382On DRACO, the same changes cut a single agent's time by 69% and its cost per task by 49%. Its score was 1.9 points lower (95% interval 1.1 to 2.8 lower), and the far end of that interval, 2.8 points lower, is past the 1.5-point margin. At `medium` effort, the single agent cost 25% less per task than at `high`, took 30% less time, and scored 0.7 points lower (95% interval 0.1 to 1.3 lower). With both changes at `high` effort, the single agent took 53% less time than at `medium` effort (95% interval 42% less to 63% less), and its cost per task was 31% less (95% interval 28% less to 35% less). Its score was 1.2 points lower (95% interval 0.5 to 1.9 lower), and the far end of that interval, 1.9 points lower, is past the 1.5-point margin.
383
384On all three sets, both changes at `high` effort saved more time than `medium` effort did. On HLE and the physics set, there was no clear difference in cost, and on DRACO the cost was lower. The score was about the same on HLE. On the physics set it was 3.2 points higher, but that interval includes zero, so the difference isn't clear. So for a single agent, the clock saves more time than a lower effort level does. On DRACO, though, the single agent with both changes scored 1.2 points lower than at `medium` effort (95% interval 0.5 to 1.9 lower).
385
386**When to use it.**
387
388* Use both changes when an agent's time matters and a small score change is acceptable. On every configuration measured, they cut the time and cost per task, for teams and for single agents.
389* Check score on your own tasks before you adopt them. On DRACO, the score was 1.5 points lower for a team and 1.9 points lower for a single agent. On HLE, it was 1.7 points lower for a team and 1.1 points lower for a single agent. On the physics set, neither score change was clearly different from zero.
390* If you're already thinking about a lower effort level to save time, compare it with the clock. A single agent with both changes at `high` effort took less time than at `medium` effort: 53% less on DRACO, 25% less on HLE, and 27% less on the physics set. Its cost per task was 31% lower on DRACO, with no clear difference on HLE and the physics set.
391
392**How to add it.** Put this instruction at the start of every agent's system prompt:
393
394```text wrap
395Time matters here: do not spend time that can be avoided, and the earlier a correct result is obtained, the better. The elapsed time so far is shown before each of your turns.
396```
397
398The second sentence tells the model that the clock messages exist. The first request carries no clock, and the measured runs used exactly this wording.
399
400Then, before each request after an agent's first, append a [mid-conversation system message](https://platform.claude.com/docs/en/build-with-claude/mid-conversation-system-messages) that gives the elapsed time in whole seconds, such as `Elapsed time: 412 seconds`. Count from the start of the task, not from the start of the agent. In a team, every agent reads the same clock, so the first clock a helper sees already counts the time the team spent before the helper started. In a tool loop, put the message right after the `user` message that carries the tool results, as [Placement after tool results](https://platform.claude.com/docs/en/build-with-claude/mid-conversation-system-messages#placement-after-tool-results) shows. If you send the agent a new `user` message instead, put the clock after that message.
401
402Leave earlier clock messages where they are. Each one becomes part of the conversation history, so the cached prefix still matches on the next request (see [Combining with prompt caching](https://platform.claude.com/docs/en/build-with-claude/mid-conversation-system-messages#combining-with-prompt-caching)). Anthropic measured these plain system messages, which stay visible to the model. A [turn-scoped system message](https://platform.claude.com/docs/en/build-with-claude/mid-conversation-system-messages#turn-scoped-system-messages) would show the model only the newest clock, and Anthropic didn't measure that form.
403
404The following example runs one agent's tool loop with both changes. It adds the clock after tool results, and it handles client tools only:
405
406```python
407import time
408
409import anthropic
410
411client = anthropic.Anthropic()
412
413TIME_MATTERS = (
414 "Time matters here: do not spend time that can be avoided, and the earlier a "
415 "correct result is obtained, the better. The elapsed time so far is shown before "
416 "each of your turns."
417)
418
419
420def run_agent(task, system, tools, run_tool, started_at=None):
421 """Run one agent's tool loop. In a team, pass the lead's started_at to every helper."""
422 if started_at is None:
423 # Wall-clock seconds, so helpers in other processes can share the lead's start time.
424 started_at = time.time()
425 messages = [{"role": "user", "content": task}]
426 while True:
427 # Stream because a 128,000-token cap is too large for a non-streaming request.
428 with client.messages.stream(
429 model="claude-fable-5-1",
430 max_tokens=128000,
431 cache_control={"type": "ephemeral"},
432 system=TIME_MATTERS + "\n\n" + system,
433 tools=tools,
434 messages=messages,
435 ) as stream:
436 response = stream.get_final_message()
437 messages.append({"role": "assistant", "content": response.content})
438 if response.stop_reason != "tool_use":
439 return response
440 results = [
441 {
442 "type": "tool_result",
443 "tool_use_id": block.id,
444 "content": run_tool(block.name, block.input),
445 }
446 for block in response.content
447 if block.type == "tool_use"
448 ]
449 messages.append({"role": "user", "content": results})
450 # A system message must follow a user turn, so the clock goes after the tool results.
451 elapsed = int(time.time() - started_at)
452 messages.append(
453 {"role": "system", "content": f"Elapsed time: {elapsed} seconds"}
454 )
455```
456
457Claude Fable 5.1 supports mid-conversation system messages. The [list of supported models](https://platform.claude.com/docs/en/build-with-claude/mid-conversation-system-messages) covers the others. On a model without them, such as Claude Sonnet 5, you can put the same line in a text block after the last `tool_result` block in the `user` turn. Anthropic measured only the system-message form.
458
459On Claude Managed Agents, you can send a [`system.message` event](https://platform.claude.com/docs/en/managed-agents/events-and-streaming#sending-system-messages) with a tool result or a user message. The message applies to that turn and every later turn. So turns that follow the platform's built-in tools, such as web search, see the last clock you sent, not the current time. A `system.message` also reaches only the session's primary thread. In a [multiagent session](https://platform.claude.com/docs/en/managed-agents/multiagent-orchestration), that is the coordinator's thread, so the worker agents never see a clock that you send this way. To show the current time before every turn, and to every agent in a team, run the agent loop yourself on the Messages API.
460
365461## Combine models
366462
367463Multi-model architectures fit workloads whose task complexity varies enough that different steps are best served by different models. When your traffic mixes routine work that a smaller model handles reliably with harder steps that need frontier capability, splitting the work keeps frontier intelligence where it matters while most tokens bill at smaller-model rates. When a workload lacks that mix, because its difficulty is uniform or it is one dependent chain, a single well-tuned model is usually the better choice. Each strategy section gives the rule for telling the two cases apart.
from line 511
415511
416512This pattern saves wall-clock time when workers can run in parallel: on the corpus benchmark[8](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#refs), an episode took about 2.3 hours with the coordinator running the platform's documented limit of 25 concurrent workers, compared with 15 to 20 hours solo. It saved money in only two measured situations. On work a single model could handle alone, the same model at lower effort was cheaper every time.
417513
514When workers run in parallel, a time instruction and an elapsed-time clock can shorten the run. On DRACO[21](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#refs), a team of same-model agents with the instruction and the clock finished in 33% less time at a 54% lower cost per task, and scored 1.5 points lower. Every agent in that team had the instruction and the clock. Anthropic didn't measure the clock with lower-cost workers. On Claude Managed Agents, the clock reaches only the coordinator, so the workers never see it. Anthropic didn't measure a team in which only the coordinator has the clock. The coordinator's clock is also current only on turns that follow your own tool results or messages. [Show the model elapsed time](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#show-the-model-elapsed-time) has the recipe for an agent loop that you run on the Messages API.
515
418516**Case 1: insurance against the cost tail on routine work.** A frontier model running alone occasionally spirals on a routine problem it would normally solve. Because you cannot tell in advance which those will be, a few such runs dominate the bill. A coordinator that hands routine work to a lower-cost worker caps that tail, because any spiraling now happens at worker rates.
419517
420518Anthropic measured this on a deliberately easy slice of BrowseComp[4](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#refs) (10 problems the solo model reliably solves; 50 delegated and 70 solo runs). A Claude Fable 5 coordinator with one Claude Sonnet 5 worker cost about half as much as Claude Fable 5 alone on average and about a third as much at the 90th percentile ($12 compared with $33), and the solo model's single most expensive run, at $84, was also wrong:
from line 837
73983718. **Compaction timing measurement:** The triage agent's long variant from [Trim input and context tokens](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#trim-input-and-context-tokens), run August 24, 2026, on Claude Sonnet 5 with the 5-minute cache, cost from the usage fields at list prices, five sessions per arm: a no-change arm at the default effort throughout ($0.81 per session), and two arms that start at low effort and make the same two cache-breaking changes, a switch to the default effort and one added tool, either mid-session at requests 12 and 17 ($0.95) or together on the first request after the first compaction ($0.75). A fourth arm of six sessions, run August 25, 2026, made the same two changes on the request that triggered the first compaction ($0.92 per session): that request's summarization pass wrote the 81,000-token context to the cache instead of reading it, so that pass cost $0.21 against $0.04 for the same pass in the boundary arm. Sessions first compacted at request 21 to 25 (16 of the 21 sessions at request 22), once the prompt passed the 80,000-token compaction trigger, and two no-change sessions compacted a second time near the end. The boundary arm's lower total than the no-change arm reflects its low-effort requests before the change and those second compactions rather than caching: the two arms' re-write costs differ by under a cent. The mid-session arm paid $0.23 per session in cache re-writes; the difference between the mid-session and boundary arms was $0.20 with a 95% confidence interval of $0.11 to $0.29. One mid-session session ran cheap ($0.82) after its model mis-called the search tool following compaction and got empty results; it is included, and without it the arm averages $0.98. Accuracy averaged 14.2 of 20 labels in each August 24 arm and 14.7 in the August 25 arm; cache reads were 91% of prompt tokens with no changes, 85% mid-session, 91% at the boundary, and 86% with the changes on the triggering request.
74083819. **Cache duration measurement on Claude Fable 5.1:** The same 20-issue triage job and harness as reference 16, run August 23 and August 26, 2026, on the Claude Fable 5.1 launch snapshot at its launch prices ($10 input, $12.50 5-minute write, $20 1-hour write, $0.25 cache read, $50 output per million tokens), three settings per schedule: the 5-minute cache, the 1-hour cache, and the 5-minute cache kept warm by a `max_tokens: 0` request on the unchanged prefix every 4 minutes, timed from the previous request's start (the August 23 runs sent keep-alive requests with `max_tokens: 1`; in the August 26 cells reported here, every keep-alive request refreshed the cache and billed no output). Schedules: no pauses, 10% of turns, and every turn at 6 minutes on all 20 issues, and 45-minute pauses on the 5-issue subset; three runs per cell (six for the August 26 keep-alive cell with 45-minute pauses), cost computed from each response's `usage` fields at list prices, accuracy against the same gold labels (12 to 17 exact labels of 20). Per-session means on August 26 for the 5-minute, 1-hour, and keep-alive settings: no pauses $2.42, $3.09, $2.29; 10% paused $4.50, $2.96, $2.36; every turn $22.89, $3.01, $2.62; the August 23 cells agree within 6%. The 45-minute figures ($1.68, $0.59, and $0.71 per 5-issue session) are from a clean re-run on August 26 after a cache-billing incident spoiled that day's first cells; the August 23 runs gave $1.67, $0.58, and $0.70. The crossover between the 5-minute and 1-hour settings is 3.1% of turns, the same measure as reference 16.
74183920. **Terminal-Bench 3:** the public terminal-agent benchmark's 74 tasks, run on [Claude Managed Agents](https://platform.claude.com/docs/en/managed-agents/overview) with two custom tools, a shell and a file editor that the evaluation harness runs in each task's own container, in place of the platform's built-in tools, and otherwise at the platform's default settings for external accounts, two runs per model at `high` effort, August 27 to 28, 2026. These runs used Terminal-Bench version 3.0, and their scores are not comparable with the public Terminal-Bench leaderboard or with the Terminal-Bench 4.0 results in the Claude Opus 5.5 system card, which come from runs in Claude Code at `max` effort. Each task's time limits are 2.5 times the benchmark's own, which gives the agent between 75 minutes and 20 hours per task (5 hours for the median task), and each task gets three times the memory it specifies, from 6 GiB to 96 GiB, with extra memory for the 12 tasks that run helper services. The agent had no general internet access: its containers could reach an internal package mirror, a short list of download sites including GitHub and the Python Package Index, and a few sites specific to some tasks, and eight of the tasks had no network access at all. Scores are raw pass rates over the 148 attempts per model; single runs swing by 5 to 11 points. Costs are what a customer would be billed at list prices, re-priced request by request from the runs' usage records with the 5-minute cache lifetime. Claude Opus 4.7 ended 11 of its 148 attempts at its output cap.
84021. **DRACO:** Perplexity, "DRACO: a Cross-Domain Benchmark for Deep Research Accuracy, Completeness, and Objectivity," arXiv:2602.11685, 2026. Its 100 research tasks across 10 domains are graded against expert-written rubrics, and the score is the benchmark's normalized score. Every configuration ran on the Claude API with Claude Fable 5.1, the default adaptive thinking, the production safety classifiers on, and `max_tokens` at 128,000: a single agent at `high` and at `medium` effort, the single agent at `high` with the instruction and the clock, and a team at `high` with and without them. The team is a lead agent that starts helper agents of the same model through a tool, with no cap on their number. On DRACO, the lead started a median of 4 helpers per attempt. Each configuration made three attempts at each task, run September 8 to 10, 2026. An attempt that hit the four-hour limit was run again, and the new attempt counts. The only attempts left out are all 3 attempts at one task for the single agent at `medium` effort, so that configuration covers 99 tasks. That task timed out on every attempt, in the original run and in the re-run. Scoring those 3 attempts as 0, as the benchmark's own scoring would, affects only the two comparisons with `medium` effort. The score change at `medium` effort against `high` moves from 0.7 to 1.7 points lower, and the score change with both changes against `medium` effort moves from 1.2 to 0.2 points lower. The agents used a search tool and a fetch tool that the evaluation harness hosts over a pinned web index. Those tools set part of the time, and yours will run at a different speed, so the page gives time as a ratio between configurations, not in minutes. Time is the wall-clock time per task, from the first request to the last request on any agent, minus the estimated time spent waiting to retry requests after rate-limit or overload errors. Those errors came from the test account's shared limits. All configurations of a set started together. The slower ones finished hours later, so part of their time ran under different load. Each task's cost is its requests priced at public list prices, with prompt caching billed as it would be for a customer who sets a cache breakpoint at the end of each request and uses the 5-minute cache lifetime, for model tokens only. The harness's tools add no charges. Score changes are paired differences over tasks, with 95% bootstrap intervals. A change counts as inside the margin when its interval stays within 1.5 points on DRACO and 2.5 points on HLE. Anthropic set those margins before the runs. Claude Opus 5 grades the answers. Against each set's own grader, Opus 5 scored 1.9 to 2.4 points higher on DRACO, 2.2 to 2.9 points lower on HLE (Opus 5 graded 495 of the 500 questions, and the benchmark's grader graded all 500), and 1.3 to 2.0 points lower on the physics set, whose own grader also uses the expert reference solutions, in every configuration. The two graders agree on the direction of every change.
84122. **HLE:** Phan et al., "Humanity's Last Exam," arXiv:2501.14249, 2025. Expert-written questions with exact answers, graded against the reference answers. Measured on the first 500 questions, with the benchmark's own sources blocked from search, and the same setup as reference 21. Each configuration made three attempts at each question, run September 8 to 10, 2026. Claude Opus 5 compares each answer with the reference answer, with adaptive thinking on, as it is by default. The judge graded 495 of the 500 questions in every configuration, and the scores cover those 495. For the other 5, the grading request was over the judge's 1M-token limit. An attempt that hit the four-hour limit was run again, and the new attempt counts, so every configuration has all 1,500 attempts. Scoring the 5 ungraded questions as 0, as the benchmark's own scoring would, changes no finding.
84223. **Physics set:** An internal set of 70 research-level physics problems, adapted from the public CritPt benchmark: Zhu et al., "Probing the Critical Point (CritPt) of AI Reasoning: a Frontier Physics Research Benchmark," arXiv:2509.26574, 2025. Expert reviewers corrected the problem statements. Claude Opus 5 grades each answer against expert reference solutions that aren't public, so the scores can't be compared with published results. The score is the mean grade over a problem's attempts, averaged over problems. Measured on all 70 problems, four attempts per problem, run September 8 to 9, 2026. Every agent had a Python tool, a shell, and a file editor in a sandbox container with no network access, and no search or fetch tools. Otherwise the setup is that of reference 21. No score margin was set for the physics set before the runs, so the page gives its score changes with their 95% intervals and doesn't describe them as inside a margin.
742843
743844## Next steps
744845
agents-and-tools/tool-use/tool-runner Changed · +3 / -3 lines
from line 589
589589
590590The tool runner is an iterable that yields messages from Claude. On each iteration, the runner checks whether Claude requested a tool use. If so, it runs the tool and sends the result back to Claude automatically, then yields the next message from Claude to continue your loop.
591591
592You can end the loop at any iteration with a `break` statement. The runner loops until Claude returns a message without a tool use, or until it reaches `max_iterations` if you set it.
592You can end the loop at any iteration with a `break` statement. The runner loops until Claude returns a message without a tool use, or until it reaches `max_iterations` (csharp, java, php: `maxIterations`; go: `MaxIterations`) if you set it.
593593
594594If you don't need intermediate messages, you can get the final message directly:
595595
from line 821
821821
822822You take over by modifying the runner's messages from inside the loop body. The exact method depends on the SDK. See the per-language tabs that follow.
823823
824When you take over for an iteration, the runner does not append the assistant message or tool results from that turn. You become responsible for keeping the conversation valid: append the assistant message and a tool result yourself (if you want the turn to count), modify state conditionally so the loop can still exit when there are no tool calls, and pass `max_iterations` to bound the loop. All seven SDKs support `max_iterations`.
824When you take over for an iteration, the runner does not append the assistant message or tool results from that turn. You become responsible for keeping the conversation valid: append the assistant message and a tool result yourself (if you want the turn to count), modify state conditionally so the loop can still exit when there are no tool calls, and pass `max_iterations` (csharp, java, php: `maxIterations`; go: `MaxIterations`) to bound the loop. All seven SDKs support `max_iterations` (csharp, java, php: `maxIterations`; go: `MaxIterations`).
825825
826826<Tabs>
827827 <Tab title="Python">
from line 1106
11061106
11071107### Automatic context management
11081108
1109For long-running agentic tasks, the TypeScript and Ruby tool runners support automatic [compaction](https://platform.claude.com/docs/en/build-with-claude/context-editing#client-side-compaction-sdk), which generates summaries when token usage exceeds a threshold so the conversation can continue beyond context window limits. Both SDKs have deprecated this client-side option in favor of [server-side compaction](https://platform.claude.com/docs/en/build-with-claude/compaction-threshold), which works with every SDK's tool runner through the `context_management` request parameter. The Python SDK (v1.0 and later) and the Go, Java, C#, and PHP tool runners don't include client-side compaction. The Python, TypeScript, C#, Go, Java, PHP, and Ruby tool runners have a `compact_before_next_turn()` helper for on-demand compaction, spelled in each language's own casing. See [Compact in a loop](https://platform.claude.com/docs/en/build-with-claude/compaction-on-demand#compact-in-a-loop). Use it or a `context_management` compaction edit on a runner, not both.
1109For long-running agentic tasks, the TypeScript and Ruby tool runners support automatic [compaction](https://platform.claude.com/docs/en/build-with-claude/context-editing#client-side-compaction-sdk), which generates summaries when token usage exceeds a threshold so the conversation can continue beyond context window limits. Both SDKs have deprecated this client-side option in favor of [server-side compaction](https://platform.claude.com/docs/en/build-with-claude/compaction-threshold), which works with every SDK's tool runner through the `context_management` request parameter. The Python SDK (v1.0 and later) and the Go, Java, C#, and PHP tool runners don't include client-side compaction. The Python, TypeScript, C#, Go, Java, PHP, and Ruby tool runners have a `compact_before_next_turn()` (typescript, java, php: `compactBeforeNextTurn()`; csharp, go: `CompactBeforeNextTurn()`) helper for on-demand compaction. See [Compact in a loop](https://platform.claude.com/docs/en/build-with-claude/compaction-on-demand#compact-in-a-loop). Use it or a `context_management` compaction edit on a runner, not both.
11101110
11111111### Debugging tool execution
11121112
api/beta Changed · +85 / -46 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.
Nothing in the body moved in this read. What changed is above.
api/beta/dreams Changed · +15 / -5 lines
from line 235
235235
236236 An asynchronous job that reads a memory store and past sessions, then writes a reorganized version of that memory store.
237237
238 By default the dream writes its result to a new memory store and doesn't change the input memory store. With `output_behavior` set to `update_existing`, it writes its result into the input memory store instead. The Dreams API is in research preview, so this resource can still change.
238 By default the dream writes its result to a new memory store and doesn't change the input memory store. With `output_behavior` set to `update_existing`, it writes its result into the input memory store instead.
239239
240 The Dreams API is in research preview: the request and response shapes are volatile and may change without the deprecation period that applies to generally-available endpoints.
241
240242 See the [Dreams guide](https://platform.claude.com/docs/en/managed-agents/dreams#how-it-works) for what a dream reads and produces.
241243
242244 - `type: "dream"`
from line 1122
11201122
11211123 An asynchronous job that reads a memory store and past sessions, then writes a reorganized version of that memory store.
11221124
1123 By default the dream writes its result to a new memory store and doesn't change the input memory store. With `output_behavior` set to `update_existing`, it writes its result into the input memory store instead. The Dreams API is in research preview, so this resource can still change.
1125 By default the dream writes its result to a new memory store and doesn't change the input memory store. With `output_behavior` set to `update_existing`, it writes its result into the input memory store instead.
11241126
1127 The Dreams API is in research preview: the request and response shapes are volatile and may change without the deprecation period that applies to generally-available endpoints.
1128
11251129 See the [Dreams guide](https://platform.claude.com/docs/en/managed-agents/dreams#how-it-works) for what a dream reads and produces.
11261130
11271131 - `type: "dream"`
from line 1529
15251529
15261530 An asynchronous job that reads a memory store and past sessions, then writes a reorganized version of that memory store.
15271531
1528 By default the dream writes its result to a new memory store and doesn't change the input memory store. With `output_behavior` set to `update_existing`, it writes its result into the input memory store instead. The Dreams API is in research preview, so this resource can still change.
1532 By default the dream writes its result to a new memory store and doesn't change the input memory store. With `output_behavior` set to `update_existing`, it writes its result into the input memory store instead.
15291533
1534 The Dreams API is in research preview: the request and response shapes are volatile and may change without the deprecation period that applies to generally-available endpoints.
1535
15301536 See the [Dreams guide](https://platform.claude.com/docs/en/managed-agents/dreams#how-it-works) for what a dream reads and produces.
15311537
15321538 - `type: "dream"`
from line 1937
19311937
19321938 An asynchronous job that reads a memory store and past sessions, then writes a reorganized version of that memory store.
19331939
1934 By default the dream writes its result to a new memory store and doesn't change the input memory store. With `output_behavior` set to `update_existing`, it writes its result into the input memory store instead. The Dreams API is in research preview, so this resource can still change.
1940 By default the dream writes its result to a new memory store and doesn't change the input memory store. With `output_behavior` set to `update_existing`, it writes its result into the input memory store instead.
19351941
1942 The Dreams API is in research preview: the request and response shapes are volatile and may change without the deprecation period that applies to generally-available endpoints.
1943
19361944 See the [Dreams guide](https://platform.claude.com/docs/en/managed-agents/dreams#how-it-works) for what a dream reads and produces.
19371945
19381946 - `type: "dream"`
from line 2219
22112219
22122220 An asynchronous job that reads a memory store and past sessions, then writes a reorganized version of that memory store.
22132221
2214 By default the dream writes its result to a new memory store and doesn't change the input memory store. With `output_behavior` set to `update_existing`, it writes its result into the input memory store instead. The Dreams API is in research preview, so this resource can still change.
2222 By default the dream writes its result to a new memory store and doesn't change the input memory store. With `output_behavior` set to `update_existing`, it writes its result into the input memory store instead.
2223
2224 The Dreams API is in research preview: the request and response shapes are volatile and may change without the deprecation period that applies to generally-available endpoints.
22152225
22162226 See the [Dreams guide](https://platform.claude.com/docs/en/managed-agents/dreams#how-it-works) for what a dream reads and produces.
22172227
api/beta/memory_stores Changed · +30 / -30 lines
from line 2615
26152615
26162616 - `created_by: optional BetaManagedAgentsActor`
26172617
2618 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
2618 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
26192619
26202620 - `BetaManagedAgentsSessionActor object`
26212621
2622 Attribution for a write made by an agent during a session, through the mounted filesystem at `/mnt/memory/`.
2622 An agent acting during a session, for example through the session's mounted filesystem. It names the session itself, not the user or API key that started the session.
26232623
26242624 - `type: "session_actor"`
26252625
26262626 - `session_id: string`
26272627
2628 ID of the session that performed the write (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
2628 ID of the session (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
26292629
26302630 minLength: 1
26312631
26322632 - `BetaManagedAgentsAPIActor object`
26332633
2634 Attribution for a write made directly via the public API (outside of any session).
2634 A direct caller of the public API, identified by the API key that authenticated the request.
26352635
26362636 - `type: "api_actor"`
26372637
26382638 - `api_key_id: string`
26392639
2640 ID of the API key that performed the write. This identifies the key, not the secret.
2640 ID of the API key (an `apikey_...` value). This identifies the key, not the secret.
26412641
26422642 minLength: 1
26432643
26442644 - `BetaManagedAgentsUserActor object`
26452645
2646 Attribution for a write made by a human user through the Anthropic Console.
2646 A human user, for example acting through the Anthropic Console.
26472647
26482648 - `type: "user_actor"`
26492649
26502650 - `user_id: string`
26512651
2652 ID of the user who performed the write (a `user_...` value).
2652 ID of the user (a `user_...` value).
26532653
26542654 minLength: 1
26552655
26562656 - `BetaManagedAgentsServiceAccountActor object`
26572657
2658 Attribution for a write made by a workload authenticated as a service account, for example via Workload Identity Federation.
2658 A workload authenticated as a service account, for example via Workload Identity Federation.
26592659
26602660 - `type: "service_account_actor"`
26612661
26622662 - `service_account_id: string`
26632663
2664 ID of the service account that performed the write (a `svac_...` value).
2664 ID of the service account (a `svac_...` value).
26652665
26662666 minLength: 1
26672667
from line 2677
26772677
26782678 - `redacted_by: optional BetaManagedAgentsActor`
26792679
2680 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
2680 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
26812681
26822682- `next_page: optional string or null`
26832683
from line 2923
29232923
29242924 - `created_by: optional BetaManagedAgentsActor`
29252925
2926 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
2926 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
29272927
29282928 - `BetaManagedAgentsSessionActor object`
29292929
2930 Attribution for a write made by an agent during a session, through the mounted filesystem at `/mnt/memory/`.
2930 An agent acting during a session, for example through the session's mounted filesystem. It names the session itself, not the user or API key that started the session.
29312931
29322932 - `type: "session_actor"`
29332933
29342934 - `session_id: string`
29352935
2936 ID of the session that performed the write (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
2936 ID of the session (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
29372937
29382938 minLength: 1
29392939
29402940 - `BetaManagedAgentsAPIActor object`
29412941
2942 Attribution for a write made directly via the public API (outside of any session).
2942 A direct caller of the public API, identified by the API key that authenticated the request.
29432943
29442944 - `type: "api_actor"`
29452945
29462946 - `api_key_id: string`
29472947
2948 ID of the API key that performed the write. This identifies the key, not the secret.
2948 ID of the API key (an `apikey_...` value). This identifies the key, not the secret.
29492949
29502950 minLength: 1
29512951
29522952 - `BetaManagedAgentsUserActor object`
29532953
2954 Attribution for a write made by a human user through the Anthropic Console.
2954 A human user, for example acting through the Anthropic Console.
29552955
29562956 - `type: "user_actor"`
29572957
29582958 - `user_id: string`
29592959
2960 ID of the user who performed the write (a `user_...` value).
2960 ID of the user (a `user_...` value).
29612961
29622962 minLength: 1
29632963
29642964 - `BetaManagedAgentsServiceAccountActor object`
29652965
2966 Attribution for a write made by a workload authenticated as a service account, for example via Workload Identity Federation.
2966 A workload authenticated as a service account, for example via Workload Identity Federation.
29672967
29682968 - `type: "service_account_actor"`
29692969
29702970 - `service_account_id: string`
29712971
2972 ID of the service account that performed the write (a `svac_...` value).
2972 ID of the service account (a `svac_...` value).
29732973
29742974 minLength: 1
29752975
from line 2985
29852985
29862986 - `redacted_by: optional BetaManagedAgentsActor`
29872987
2988 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
2988 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
29892989
29902990#### Example
29912991
from line 3208
32083208
32093209 - `created_by: optional BetaManagedAgentsActor`
32103210
3211 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
3211 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
32123212
32133213 - `BetaManagedAgentsSessionActor object`
32143214
3215 Attribution for a write made by an agent during a session, through the mounted filesystem at `/mnt/memory/`.
3215 An agent acting during a session, for example through the session's mounted filesystem. It names the session itself, not the user or API key that started the session.
32163216
32173217 - `type: "session_actor"`
32183218
32193219 - `session_id: string`
32203220
3221 ID of the session that performed the write (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
3221 ID of the session (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
32223222
32233223 minLength: 1
32243224
32253225 - `BetaManagedAgentsAPIActor object`
32263226
3227 Attribution for a write made directly via the public API (outside of any session).
3227 A direct caller of the public API, identified by the API key that authenticated the request.
32283228
32293229 - `type: "api_actor"`
32303230
32313231 - `api_key_id: string`
32323232
3233 ID of the API key that performed the write. This identifies the key, not the secret.
3233 ID of the API key (an `apikey_...` value). This identifies the key, not the secret.
32343234
32353235 minLength: 1
32363236
32373237 - `BetaManagedAgentsUserActor object`
32383238
3239 Attribution for a write made by a human user through the Anthropic Console.
3239 A human user, for example acting through the Anthropic Console.
32403240
32413241 - `type: "user_actor"`
32423242
32433243 - `user_id: string`
32443244
3245 ID of the user who performed the write (a `user_...` value).
3245 ID of the user (a `user_...` value).
32463246
32473247 minLength: 1
32483248
32493249 - `BetaManagedAgentsServiceAccountActor object`
32503250
3251 Attribution for a write made by a workload authenticated as a service account, for example via Workload Identity Federation.
3251 A workload authenticated as a service account, for example via Workload Identity Federation.
32523252
32533253 - `type: "service_account_actor"`
32543254
32553255 - `service_account_id: string`
32563256
3257 ID of the service account that performed the write (a `svac_...` value).
3257 ID of the service account (a `svac_...` value).
32583258
32593259 minLength: 1
32603260
from line 3270
32703270
32713271 - `redacted_by: optional BetaManagedAgentsActor`
32723272
3273 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
3273 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
32743274
32753275#### Example
32763276
api/beta/memory_stores/memory_versions Changed · +57 / -57 lines
from line 257
257257
258258 - `created_by: optional BetaManagedAgentsActor`
259259
260 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
260 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
261261
262262 - `BetaManagedAgentsSessionActor object`
263263
264 Attribution for a write made by an agent during a session, through the mounted filesystem at `/mnt/memory/`.
264 An agent acting during a session, for example through the session's mounted filesystem. It names the session itself, not the user or API key that started the session.
265265
266266 - `type: "session_actor"`
267267
268268 - `session_id: string`
269269
270 ID of the session that performed the write (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
270 ID of the session (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
271271
272272 minLength: 1
273273
274274 - `BetaManagedAgentsAPIActor object`
275275
276 Attribution for a write made directly via the public API (outside of any session).
276 A direct caller of the public API, identified by the API key that authenticated the request.
277277
278278 - `type: "api_actor"`
279279
280280 - `api_key_id: string`
281281
282 ID of the API key that performed the write. This identifies the key, not the secret.
282 ID of the API key (an `apikey_...` value). This identifies the key, not the secret.
283283
284284 minLength: 1
285285
286286 - `BetaManagedAgentsUserActor object`
287287
288 Attribution for a write made by a human user through the Anthropic Console.
288 A human user, for example acting through the Anthropic Console.
289289
290290 - `type: "user_actor"`
291291
292292 - `user_id: string`
293293
294 ID of the user who performed the write (a `user_...` value).
294 ID of the user (a `user_...` value).
295295
296296 minLength: 1
297297
298298 - `BetaManagedAgentsServiceAccountActor object`
299299
300 Attribution for a write made by a workload authenticated as a service account, for example via Workload Identity Federation.
300 A workload authenticated as a service account, for example via Workload Identity Federation.
301301
302302 - `type: "service_account_actor"`
303303
304304 - `service_account_id: string`
305305
306 ID of the service account that performed the write (a `svac_...` value).
306 ID of the service account (a `svac_...` value).
307307
308308 minLength: 1
309309
from line 319
319319
320320 - `redacted_by: optional BetaManagedAgentsActor`
321321
322 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
322 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
323323
324324- `next_page: optional string or null`
325325
from line 565
565565
566566 - `created_by: optional BetaManagedAgentsActor`
567567
568 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
568 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
569569
570570 - `BetaManagedAgentsSessionActor object`
571571
572 Attribution for a write made by an agent during a session, through the mounted filesystem at `/mnt/memory/`.
572 An agent acting during a session, for example through the session's mounted filesystem. It names the session itself, not the user or API key that started the session.
573573
574574 - `type: "session_actor"`
575575
576576 - `session_id: string`
577577
578 ID of the session that performed the write (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
578 ID of the session (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
579579
580580 minLength: 1
581581
582582 - `BetaManagedAgentsAPIActor object`
583583
584 Attribution for a write made directly via the public API (outside of any session).
584 A direct caller of the public API, identified by the API key that authenticated the request.
585585
586586 - `type: "api_actor"`
587587
588588 - `api_key_id: string`
589589
590 ID of the API key that performed the write. This identifies the key, not the secret.
590 ID of the API key (an `apikey_...` value). This identifies the key, not the secret.
591591
592592 minLength: 1
593593
594594 - `BetaManagedAgentsUserActor object`
595595
596 Attribution for a write made by a human user through the Anthropic Console.
596 A human user, for example acting through the Anthropic Console.
597597
598598 - `type: "user_actor"`
599599
600600 - `user_id: string`
601601
602 ID of the user who performed the write (a `user_...` value).
602 ID of the user (a `user_...` value).
603603
604604 minLength: 1
605605
606606 - `BetaManagedAgentsServiceAccountActor object`
607607
608 Attribution for a write made by a workload authenticated as a service account, for example via Workload Identity Federation.
608 A workload authenticated as a service account, for example via Workload Identity Federation.
609609
610610 - `type: "service_account_actor"`
611611
612612 - `service_account_id: string`
613613
614 ID of the service account that performed the write (a `svac_...` value).
614 ID of the service account (a `svac_...` value).
615615
616616 minLength: 1
617617
from line 627
627627
628628 - `redacted_by: optional BetaManagedAgentsActor`
629629
630 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
630 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
631631
632632### Example
633633
from line 850
850850
851851 - `created_by: optional BetaManagedAgentsActor`
852852
853 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
853 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
854854
855855 - `BetaManagedAgentsSessionActor object`
856856
857 Attribution for a write made by an agent during a session, through the mounted filesystem at `/mnt/memory/`.
857 An agent acting during a session, for example through the session's mounted filesystem. It names the session itself, not the user or API key that started the session.
858858
859859 - `type: "session_actor"`
860860
861861 - `session_id: string`
862862
863 ID of the session that performed the write (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
863 ID of the session (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
864864
865865 minLength: 1
866866
867867 - `BetaManagedAgentsAPIActor object`
868868
869 Attribution for a write made directly via the public API (outside of any session).
869 A direct caller of the public API, identified by the API key that authenticated the request.
870870
871871 - `type: "api_actor"`
872872
873873 - `api_key_id: string`
874874
875 ID of the API key that performed the write. This identifies the key, not the secret.
875 ID of the API key (an `apikey_...` value). This identifies the key, not the secret.
876876
877877 minLength: 1
878878
879879 - `BetaManagedAgentsUserActor object`
880880
881 Attribution for a write made by a human user through the Anthropic Console.
881 A human user, for example acting through the Anthropic Console.
882882
883883 - `type: "user_actor"`
884884
885885 - `user_id: string`
886886
887 ID of the user who performed the write (a `user_...` value).
887 ID of the user (a `user_...` value).
888888
889889 minLength: 1
890890
891891 - `BetaManagedAgentsServiceAccountActor object`
892892
893 Attribution for a write made by a workload authenticated as a service account, for example via Workload Identity Federation.
893 A workload authenticated as a service account, for example via Workload Identity Federation.
894894
895895 - `type: "service_account_actor"`
896896
897897 - `service_account_id: string`
898898
899 ID of the service account that performed the write (a `svac_...` value).
899 ID of the service account (a `svac_...` value).
900900
901901 minLength: 1
902902
from line 912
912912
913913 - `redacted_by: optional BetaManagedAgentsActor`
914914
915 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
915 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
916916
917917### Example
918918
from line 956
956956
957957- `BetaManagedAgentsActor = BetaManagedAgentsSessionActor or BetaManagedAgentsAPIActor or BetaManagedAgentsUserActor or BetaManagedAgentsServiceAccountActor`
958958
959 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
959 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
960960
961961 - `BetaManagedAgentsSessionActor object`
962962
963 Attribution for a write made by an agent during a session, through the mounted filesystem at `/mnt/memory/`.
963 An agent acting during a session, for example through the session's mounted filesystem. It names the session itself, not the user or API key that started the session.
964964
965965 - `type: "session_actor"`
966966
967967 - `session_id: string`
968968
969 ID of the session that performed the write (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
969 ID of the session (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
970970
971971 minLength: 1
972972
973973 - `BetaManagedAgentsAPIActor object`
974974
975 Attribution for a write made directly via the public API (outside of any session).
975 A direct caller of the public API, identified by the API key that authenticated the request.
976976
977977 - `type: "api_actor"`
978978
979979 - `api_key_id: string`
980980
981 ID of the API key that performed the write. This identifies the key, not the secret.
981 ID of the API key (an `apikey_...` value). This identifies the key, not the secret.
982982
983983 minLength: 1
984984
985985 - `BetaManagedAgentsUserActor object`
986986
987 Attribution for a write made by a human user through the Anthropic Console.
987 A human user, for example acting through the Anthropic Console.
988988
989989 - `type: "user_actor"`
990990
991991 - `user_id: string`
992992
993 ID of the user who performed the write (a `user_...` value).
993 ID of the user (a `user_...` value).
994994
995995 minLength: 1
996996
997997 - `BetaManagedAgentsServiceAccountActor object`
998998
999 Attribution for a write made by a workload authenticated as a service account, for example via Workload Identity Federation.
999 A workload authenticated as a service account, for example via Workload Identity Federation.
10001000
10011001 - `type: "service_account_actor"`
10021002
10031003 - `service_account_id: string`
10041004
1005 ID of the service account that performed the write (a `svac_...` value).
1005 ID of the service account (a `svac_...` value).
10061006
10071007 minLength: 1
10081008
from line 1010
10101010
10111011- `BetaManagedAgentsAPIActor object`
10121012
1013 Attribution for a write made directly via the public API (outside of any session).
1013 A direct caller of the public API, identified by the API key that authenticated the request.
10141014
10151015 - `type: "api_actor"`
10161016
10171017 - `api_key_id: string`
10181018
1019 ID of the API key that performed the write. This identifies the key, not the secret.
1019 ID of the API key (an `apikey_...` value). This identifies the key, not the secret.
10201020
10211021 minLength: 1
10221022
from line 1078
10781078
10791079 - `created_by: optional BetaManagedAgentsActor`
10801080
1081 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
1081 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
10821082
10831083 - `BetaManagedAgentsSessionActor object`
10841084
1085 Attribution for a write made by an agent during a session, through the mounted filesystem at `/mnt/memory/`.
1085 An agent acting during a session, for example through the session's mounted filesystem. It names the session itself, not the user or API key that started the session.
10861086
10871087 - `type: "session_actor"`
10881088
10891089 - `session_id: string`
10901090
1091 ID of the session that performed the write (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
1091 ID of the session (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
10921092
10931093 minLength: 1
10941094
10951095 - `BetaManagedAgentsAPIActor object`
10961096
1097 Attribution for a write made directly via the public API (outside of any session).
1097 A direct caller of the public API, identified by the API key that authenticated the request.
10981098
10991099 - `type: "api_actor"`
11001100
11011101 - `api_key_id: string`
11021102
1103 ID of the API key that performed the write. This identifies the key, not the secret.
1103 ID of the API key (an `apikey_...` value). This identifies the key, not the secret.
11041104
11051105 minLength: 1
11061106
11071107 - `BetaManagedAgentsUserActor object`
11081108
1109 Attribution for a write made by a human user through the Anthropic Console.
1109 A human user, for example acting through the Anthropic Console.
11101110
11111111 - `type: "user_actor"`
11121112
11131113 - `user_id: string`
11141114
1115 ID of the user who performed the write (a `user_...` value).
1115 ID of the user (a `user_...` value).
11161116
11171117 minLength: 1
11181118
11191119 - `BetaManagedAgentsServiceAccountActor object`
11201120
1121 Attribution for a write made by a workload authenticated as a service account, for example via Workload Identity Federation.
1121 A workload authenticated as a service account, for example via Workload Identity Federation.
11221122
11231123 - `type: "service_account_actor"`
11241124
11251125 - `service_account_id: string`
11261126
1127 ID of the service account that performed the write (a `svac_...` value).
1127 ID of the service account (a `svac_...` value).
11281128
11291129 minLength: 1
11301130
from line 1140
11401140
11411141 - `redacted_by: optional BetaManagedAgentsActor`
11421142
1143 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
1143 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
11441144
11451145### Beta Managed Agents Memory Version Operation
11461146
from line 1164
11641164
11651165- `BetaManagedAgentsServiceAccountActor object`
11661166
1167 Attribution for a write made by a workload authenticated as a service account, for example via Workload Identity Federation.
1167 A workload authenticated as a service account, for example via Workload Identity Federation.
11681168
11691169 - `type: "service_account_actor"`
11701170
11711171 - `service_account_id: string`
11721172
1173 ID of the service account that performed the write (a `svac_...` value).
1173 ID of the service account (a `svac_...` value).
11741174
11751175 minLength: 1
11761176
from line 1178
11781178
11791179- `BetaManagedAgentsSessionActor object`
11801180
1181 Attribution for a write made by an agent during a session, through the mounted filesystem at `/mnt/memory/`.
1181 An agent acting during a session, for example through the session's mounted filesystem. It names the session itself, not the user or API key that started the session.
11821182
11831183 - `type: "session_actor"`
11841184
11851185 - `session_id: string`
11861186
1187 ID of the session that performed the write (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
1187 ID of the session (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
11881188
11891189 minLength: 1
11901190
from line 1192
11921192
11931193- `BetaManagedAgentsUserActor object`
11941194
1195 Attribution for a write made by a human user through the Anthropic Console.
1195 A human user, for example acting through the Anthropic Console.
11961196
11971197 - `type: "user_actor"`
11981198
11991199 - `user_id: string`
12001200
1201 ID of the user who performed the write (a `user_...` value).
1201 ID of the user (a `user_...` value).
12021202
12031203 minLength: 1
12041204
api/beta/memory_stores/memory_versions/list Changed · +10 / -10 lines
from line 255
255255
256256 - `created_by: optional BetaManagedAgentsActor`
257257
258 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
258 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
259259
260260 - `BetaManagedAgentsSessionActor object`
261261
262 Attribution for a write made by an agent during a session, through the mounted filesystem at `/mnt/memory/`.
262 An agent acting during a session, for example through the session's mounted filesystem. It names the session itself, not the user or API key that started the session.
263263
264264 - `type: "session_actor"`
265265
266266 - `session_id: string`
267267
268 ID of the session that performed the write (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
268 ID of the session (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
269269
270270 minLength: 1
271271
272272 - `BetaManagedAgentsAPIActor object`
273273
274 Attribution for a write made directly via the public API (outside of any session).
274 A direct caller of the public API, identified by the API key that authenticated the request.
275275
276276 - `type: "api_actor"`
277277
278278 - `api_key_id: string`
279279
280 ID of the API key that performed the write. This identifies the key, not the secret.
280 ID of the API key (an `apikey_...` value). This identifies the key, not the secret.
281281
282282 minLength: 1
283283
284284 - `BetaManagedAgentsUserActor object`
285285
286 Attribution for a write made by a human user through the Anthropic Console.
286 A human user, for example acting through the Anthropic Console.
287287
288288 - `type: "user_actor"`
289289
290290 - `user_id: string`
291291
292 ID of the user who performed the write (a `user_...` value).
292 ID of the user (a `user_...` value).
293293
294294 minLength: 1
295295
296296 - `BetaManagedAgentsServiceAccountActor object`
297297
298 Attribution for a write made by a workload authenticated as a service account, for example via Workload Identity Federation.
298 A workload authenticated as a service account, for example via Workload Identity Federation.
299299
300300 - `type: "service_account_actor"`
301301
302302 - `service_account_id: string`
303303
304 ID of the service account that performed the write (a `svac_...` value).
304 ID of the service account (a `svac_...` value).
305305
306306 minLength: 1
307307
from line 317
317317
318318 - `redacted_by: optional BetaManagedAgentsActor`
319319
320 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
320 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
321321
322322- `next_page: optional string or null`
323323
api/beta/memory_stores/memory_versions/redact Changed · +10 / -10 lines
from line 189
189189
190190 - `created_by: optional BetaManagedAgentsActor`
191191
192 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
192 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
193193
194194 - `BetaManagedAgentsSessionActor object`
195195
196 Attribution for a write made by an agent during a session, through the mounted filesystem at `/mnt/memory/`.
196 An agent acting during a session, for example through the session's mounted filesystem. It names the session itself, not the user or API key that started the session.
197197
198198 - `type: "session_actor"`
199199
200200 - `session_id: string`
201201
202 ID of the session that performed the write (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
202 ID of the session (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
203203
204204 minLength: 1
205205
206206 - `BetaManagedAgentsAPIActor object`
207207
208 Attribution for a write made directly via the public API (outside of any session).
208 A direct caller of the public API, identified by the API key that authenticated the request.
209209
210210 - `type: "api_actor"`
211211
212212 - `api_key_id: string`
213213
214 ID of the API key that performed the write. This identifies the key, not the secret.
214 ID of the API key (an `apikey_...` value). This identifies the key, not the secret.
215215
216216 minLength: 1
217217
218218 - `BetaManagedAgentsUserActor object`
219219
220 Attribution for a write made by a human user through the Anthropic Console.
220 A human user, for example acting through the Anthropic Console.
221221
222222 - `type: "user_actor"`
223223
224224 - `user_id: string`
225225
226 ID of the user who performed the write (a `user_...` value).
226 ID of the user (a `user_...` value).
227227
228228 minLength: 1
229229
230230 - `BetaManagedAgentsServiceAccountActor object`
231231
232 Attribution for a write made by a workload authenticated as a service account, for example via Workload Identity Federation.
232 A workload authenticated as a service account, for example via Workload Identity Federation.
233233
234234 - `type: "service_account_actor"`
235235
236236 - `service_account_id: string`
237237
238 ID of the service account that performed the write (a `svac_...` value).
238 ID of the service account (a `svac_...` value).
239239
240240 minLength: 1
241241
from line 251
251251
252252 - `redacted_by: optional BetaManagedAgentsActor`
253253
254 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
254 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
255255
256256## Example
257257
api/beta/memory_stores/memory_versions/retrieve Changed · +10 / -10 lines
from line 203
203203
204204 - `created_by: optional BetaManagedAgentsActor`
205205
206 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
206 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
207207
208208 - `BetaManagedAgentsSessionActor object`
209209
210 Attribution for a write made by an agent during a session, through the mounted filesystem at `/mnt/memory/`.
210 An agent acting during a session, for example through the session's mounted filesystem. It names the session itself, not the user or API key that started the session.
211211
212212 - `type: "session_actor"`
213213
214214 - `session_id: string`
215215
216 ID of the session that performed the write (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
216 ID of the session (a `sesn_...` value). Look up the session via [Retrieve a session](/docs/en/api/beta/sessions/retrieve) for further provenance.
217217
218218 minLength: 1
219219
220220 - `BetaManagedAgentsAPIActor object`
221221
222 Attribution for a write made directly via the public API (outside of any session).
222 A direct caller of the public API, identified by the API key that authenticated the request.
223223
224224 - `type: "api_actor"`
225225
226226 - `api_key_id: string`
227227
228 ID of the API key that performed the write. This identifies the key, not the secret.
228 ID of the API key (an `apikey_...` value). This identifies the key, not the secret.
229229
230230 minLength: 1
231231
232232 - `BetaManagedAgentsUserActor object`
233233
234 Attribution for a write made by a human user through the Anthropic Console.
234 A human user, for example acting through the Anthropic Console.
235235
236236 - `type: "user_actor"`
237237
238238 - `user_id: string`
239239
240 ID of the user who performed the write (a `user_...` value).
240 ID of the user (a `user_...` value).
241241
242242 minLength: 1
243243
244244 - `BetaManagedAgentsServiceAccountActor object`
245245
246 Attribution for a write made by a workload authenticated as a service account, for example via Workload Identity Federation.
246 A workload authenticated as a service account, for example via Workload Identity Federation.
247247
248248 - `type: "service_account_actor"`
249249
250250 - `service_account_id: string`
251251
252 ID of the service account that performed the write (a `svac_...` value).
252 ID of the service account (a `svac_...` value).
253253
254254 minLength: 1
255255
from line 265
265265
266266 - `redacted_by: optional BetaManagedAgentsActor`
267267
268 Identifies who performed a write or redact operation. Captured at write time on the `memory_version` row. The API key that created a session is not recorded on agent writes; attribution answers who made the write, not who is ultimately responsible. Look up session provenance separately via the [Sessions API](/docs/en/api/beta/sessions/retrieve).
268 Identifies who performed an operation. Recorded when the operation happens and not updated afterwards, so the ID may refer to a user, service account, API key, or session that has since been deleted.
269269
270270## Example
271271
api/beta/messages Changed · +64 / -8 lines
### Beta Cache Miss Reason
This page is larger than the 256 KiB this site keeps, so one side of the diff below stops where the stored text does.
Nothing in the body moved in this read. What changed is above.
api/beta/messages/batches Changed · +4 / -4 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.
Nothing in the body moved in this read. What changed is above.
api/beta/organization Changed · +41 / -10 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 9042
90429042
90439043**GET** `/v1/organizations/workspaces/{workspace_id}/rate_limits`
90449044
9045List rate-limit overrides configured for a workspace.
9045List a workspace's rate limits.
90469046
9047Returns only the groups and limiter types that have a workspace-level
9048override. Groups without overrides inherit the organization limits and
9049are not listed; use `GET /v1/organizations/rate_limits` to see those.
9047By default, returns only the groups and limiter types that have a
9048workspace-level override. With `include_inherited=true`, returns every
9049group with organization-level limits the workspace can see, listing for
9050each the values it inherits from the organization as well as its own
9051overrides. Each value's `source` says which it is.
90509052
90519053When `limit` is omitted, every matching entry is returned in a single
90529054page; when `limit` truncates the result, follow `next_page` to fetch
from line 9078
90769078
90779079 - `"web_search"`
90789080
9081- `include_inherited: optional boolean`
9082
9083 Also list the limiter values the workspace inherits from the organization, including groups with no workspace-level override.
9084
9085 default: false
9086
90799087- `limit: optional number`
90809088
90819089 Maximum number of items to return per page. Ranges from `1` to `1000`.
from line 9100
90929100
90939101- `data: array of BetaWorkspaceRateLimit`
90949102
9095 Rate-limit entries for the workspace, one per group that has at least one override.
9103 Rate-limit entries for the workspace: one per group with at least one override, or, with `include_inherited` set to `true`, one per group the workspace can see that has organization-level limits.
90969104
90979105 - `type: "workspace_rate_limit"`
90989106
from line 9190
91829190
91839191 - `limits: array of BetaWorkspaceRateLimitValue`
91849192
9185 The limiter values overridden for this group in this workspace. Limiter types without a workspace override are omitted and inherit the organization value.
9193 The workspace's limiter values for this group. By default only the limiter types with a workspace-level override are listed. With `include_inherited` set to `true`, the limiter types the workspace inherits from the organization are listed too, each marked by `source`.
91869194
91879195 - `type: string`
91889196
from line 9200
91929200
91939201 The organization-level value for the same limiter type, for reference. `null` when the organization has no limit configured for this limiter type.
91949202
9203 - `source: BetaWorkspaceRateLimitWorkspaceSource or BetaWorkspaceRateLimitOrganizationSource`
9204
9205 Where `value` comes from. `organization` values are listed only when `include_inherited` is `true`, and then `value` equals `org_limit`.
9206
9207 - `BetaWorkspaceRateLimitWorkspaceSource object`
9208
9209 - `type: "workspace"`
9210
9211 Always `workspace`: a workspace-level override is stored.
9212
9213 default: workspace
9214
9215 - `BetaWorkspaceRateLimitOrganizationSource object`
9216
9217 - `type: "organization"`
9218
9219 Always `organization`: no workspace-level override is stored, so the organization's value applies.
9220
9221 default: organization
9222
91959223 - `value: number`
91969224
9197 The workspace-level override value for this limiter type.
9225 The workspace's value for this limiter type: the workspace-level override when `source.type` is `workspace`, otherwise the organization's value.
91989226
91999227 - `models: array of string or null`
92009228
from line 9230
92029230
92039231 - `rate_limit_id: string`
92049232
9205 The `id` of the organization's RateLimit entry this override applies to.
9233 The `id` of the organization's RateLimit entry this entry applies to.
92069234
92079235 - `workspace_id: string`
92089236
9209 ID of the Workspace this override applies to.
9237 ID of the Workspace this entry applies to.
92109238
92119239 - `group_type: "batch" or "files" or "model_group" or 3 more`
92129240
from line 9281
92539281 "limits": [
92549282 {
92559283 "org_limit": 0,
9284 "source": {
9285 "type": "workspace"
9286 },
92569287 "type": "type",
92579288 "value": 0
92589289 }
from line 10530
1049910530Remove a service account from a workspace.
1050010531
1050110532Removal is idempotent (returns 200 even if the membership was already
10502removed). A DELETE against the implicit default-workspace membership
10503returns 200 but is a no-op and the membership persists; deleting an
10504explicit default-workspace row reverts to the implicit `workspace_user`
10505membership. Archived workspaces return 400.
10506
10507#### Path parameters
10508
10509- `workspace_id: string`
10510
10511 ID of the workspace.
10512
10513- `service_account_id: string`
10514
10515 ID of the service account.
10516
10517#### Headers
10518
10519- `"anthropic-beta": optional array of AnthropicBeta`
10520
10521 Optional header to specify the beta version(s) you want to use.
10522
10523 - `string`
10524
10525 - `"message-batches-2024-09-24" or "prompt-caching-2024-07-31" or "computer-use-2024-10-22" or 45 more`
10526
10527 - `"message-batches-2024-09-24"`
10528
10529 - `"prompt-caching-2024-07-31"`
10530
10531 - `"computer-use-2024-10-22"`
10532
10533 - `"computer-use-2025-01-24"`
10534
10535 - `"pdfs-2024-09-25"`
10536
10537 - `"token-counting-2024-11-01"`
10538
10539 - `"token-efficient-tools-2025-02-19"`
10540
10541 - `"output-128k-2025-02-19"`
10542
10543 - `"files-api-2025-04-14"`
10544
10545 - `"mcp-client-2025-04-04"`
10546
10547 - `"mcp-client-2025-11-20"`
10548
10549 - `"dev-full-thinking-2025-05-14"`
10550
10551 - `"interleaved-thinking-2025-05-14"`
10552
10553 - `"code-execution-2025-05-22"`
10554
10555 - `"extended-cache-ttl-2025-04-11"`
10556
10557 - `"context-1m-2025-08-07"`
10558
10559 - `"context-management-2025-06-27"`
10560
10561 - `"model-context-window-exceeded-2025-08-26"`
10562
10563 - `"skills-2025-10-02"`
10564
10565 - `
10533removed). A DELETE against the
api/beta/organization/workspaces Changed · +41 / -10 lines
from line 1226
12261226
12271227**GET** `/v1/organizations/workspaces/{workspace_id}/rate_limits`
12281228
1229List rate-limit overrides configured for a workspace.
1229List a workspace's rate limits.
12301230
1231Returns only the groups and limiter types that have a workspace-level
1232override. Groups without overrides inherit the organization limits and
1233are not listed; use `GET /v1/organizations/rate_limits` to see those.
1231By default, returns only the groups and limiter types that have a
1232workspace-level override. With `include_inherited=true`, returns every
1233group with organization-level limits the workspace can see, listing for
1234each the values it inherits from the organization as well as its own
1235overrides. Each value's `source` says which it is.
12341236
12351237When `limit` is omitted, every matching entry is returned in a single
12361238page; when `limit` truncates the result, follow `next_page` to fetch
from line 1262
12601262
12611263 - `"web_search"`
12621264
1265- `include_inherited: optional boolean`
1266
1267 Also list the limiter values the workspace inherits from the organization, including groups with no workspace-level override.
1268
1269 default: false
1270
12631271- `limit: optional number`
12641272
12651273 Maximum number of items to return per page. Ranges from `1` to `1000`.
from line 1284
12761284
12771285- `data: array of BetaWorkspaceRateLimit`
12781286
1279 Rate-limit entries for the workspace, one per group that has at least one override.
1287 Rate-limit entries for the workspace: one per group with at least one override, or, with `include_inherited` set to `true`, one per group the workspace can see that has organization-level limits.
12801288
12811289 - `type: "workspace_rate_limit"`
12821290
from line 1374
13661374
13671375 - `limits: array of BetaWorkspaceRateLimitValue`
13681376
1369 The limiter values overridden for this group in this workspace. Limiter types without a workspace override are omitted and inherit the organization value.
1377 The workspace's limiter values for this group. By default only the limiter types with a workspace-level override are listed. With `include_inherited` set to `true`, the limiter types the workspace inherits from the organization are listed too, each marked by `source`.
13701378
13711379 - `type: string`
13721380
from line 1384
13761384
13771385 The organization-level value for the same limiter type, for reference. `null` when the organization has no limit configured for this limiter type.
13781386
1387 - `source: BetaWorkspaceRateLimitWorkspaceSource or BetaWorkspaceRateLimitOrganizationSource`
1388
1389 Where `value` comes from. `organization` values are listed only when `include_inherited` is `true`, and then `value` equals `org_limit`.
1390
1391 - `BetaWorkspaceRateLimitWorkspaceSource object`
1392
1393 - `type: "workspace"`
1394
1395 Always `workspace`: a workspace-level override is stored.
1396
1397 default: workspace
1398
1399 - `BetaWorkspaceRateLimitOrganizationSource object`
1400
1401 - `type: "organization"`
1402
1403 Always `organization`: no workspace-level override is stored, so the organization's value applies.
1404
1405 default: organization
1406
13791407 - `value: number`
13801408
1381 The workspace-level override value for this limiter type.
1409 The workspace's value for this limiter type: the workspace-level override when `source.type` is `workspace`, otherwise the organization's value.
13821410
13831411 - `models: array of string or null`
13841412
from line 1414
13861414
13871415 - `rate_limit_id: string`
13881416
1389 The `id` of the organization's RateLimit entry this override applies to.
1417 The `id` of the organization's RateLimit entry this entry applies to.
13901418
13911419 - `workspace_id: string`
13921420
1393 ID of the Workspace this override applies to.
1421 ID of the Workspace this entry applies to.
13941422
13951423 - `group_type: "batch" or "files" or "model_group" or 3 more`
13961424
from line 1465
14371465 "limits": [
14381466 {
14391467 "org_limit": 0,
1468 "source": {
1469 "type": "workspace"
1470 },
14401471 "type": "type",
14411472 "value": 0
14421473 }
api/beta/organization/workspaces/rate_limits Changed · +106 / -15 lines
### Beta Workspace Rate Limit Organization Source ### Beta Workspace Rate Limit Workspace Source
from line 9
99
1010**GET** `/v1/organizations/workspaces/{workspace_id}/rate_limits`
1111
12List rate-limit overrides configured for a workspace.
12List a workspace's rate limits.
1313
14Returns only the groups and limiter types that have a workspace-level
15override. Groups without overrides inherit the organization limits and
16are not listed; use `GET /v1/organizations/rate_limits` to see those.
14By default, returns only the groups and limiter types that have a
15workspace-level override. With `include_inherited=true`, returns every
16group with organization-level limits the workspace can see, listing for
17each the values it inherits from the organization as well as its own
18overrides. Each value's `source` says which it is.
1719
1820When `limit` is omitted, every matching entry is returned in a single
1921page; when `limit` truncates the result, follow `next_page` to fetch
from line 45
4345
4446 - `"web_search"`
4547
48- `include_inherited: optional boolean`
49
50 Also list the limiter values the workspace inherits from the organization, including groups with no workspace-level override.
51
52 default: false
53
4654- `limit: optional number`
4755
4856 Maximum number of items to return per page. Ranges from `1` to `1000`.
from line 67
5967
6068- `data: array of BetaWorkspaceRateLimit`
6169
62 Rate-limit entries for the workspace, one per group that has at least one override.
70 Rate-limit entries for the workspace: one per group with at least one override, or, with `include_inherited` set to `true`, one per group the workspace can see that has organization-level limits.
6371
6472 - `type: "workspace_rate_limit"`
6573
from line 157
149157
150158 - `limits: array of BetaWorkspaceRateLimitValue`
151159
152 The limiter values overridden for this group in this workspace. Limiter types without a workspace override are omitted and inherit the organization value.
160 The workspace's limiter values for this group. By default only the limiter types with a workspace-level override are listed. With `include_inherited` set to `true`, the limiter types the workspace inherits from the organization are listed too, each marked by `source`.
153161
154162 - `type: string`
155163
from line 167
159167
160168 The organization-level value for the same limiter type, for reference. `null` when the organization has no limit configured for this limiter type.
161169
170 - `source: BetaWorkspaceRateLimitWorkspaceSource or BetaWorkspaceRateLimitOrganizationSource`
171
172 Where `value` comes from. `organization` values are listed only when `include_inherited` is `true`, and then `value` equals `org_limit`.
173
174 - `BetaWorkspaceRateLimitWorkspaceSource object`
175
176 - `type: "workspace"`
177
178 Always `workspace`: a workspace-level override is stored.
179
180 default: workspace
181
182 - `BetaWorkspaceRateLimitOrganizationSource object`
183
184 - `type: "organization"`
185
186 Always `organization`: no workspace-level override is stored, so the organization's value applies.
187
188 default: organization
189
162190 - `value: number`
163191
164 The workspace-level override value for this limiter type.
192 The workspace's value for this limiter type: the workspace-level override when `source.type` is `workspace`, otherwise the organization's value.
165193
166194 - `models: array of string or null`
167195
from line 197
169197
170198 - `rate_limit_id: string`
171199
172 The `id` of the organization's RateLimit entry this override applies to.
200 The `id` of the organization's RateLimit entry this entry applies to.
173201
174202 - `workspace_id: string`
175203
176 ID of the Workspace this override applies to.
204 ID of the Workspace this entry applies to.
177205
178206 - `group_type: "batch" or "files" or "model_group" or 3 more`
179207
from line 248
220248 "limits": [
221249 {
222250 "org_limit": 0,
251 "source": {
252 "type": "workspace"
253 },
223254 "type": "type",
224255 "value": 0
225256 }
from line 361
330361
331362 - `limits: array of BetaWorkspaceRateLimitValue`
332363
333 The limiter values overridden for this group in this workspace. Limiter types without a workspace override are omitted and inherit the organization value.
364 The workspace's limiter values for this group. By default only the limiter types with a workspace-level override are listed. With `include_inherited` set to `true`, the limiter types the workspace inherits from the organization are listed too, each marked by `source`.
334365
335366 - `type: string`
336367
from line 371
340371
341372 The organization-level value for the same limiter type, for reference. `null` when the organization has no limit configured for this limiter type.
342373
374 - `source: BetaWorkspaceRateLimitWorkspaceSource or BetaWorkspaceRateLimitOrganizationSource`
375
376 Where `value` comes from. `organization` values are listed only when `include_inherited` is `true`, and then `value` equals `org_limit`.
377
378 - `BetaWorkspaceRateLimitWorkspaceSource object`
379
380 - `type: "workspace"`
381
382 Always `workspace`: a workspace-level override is stored.
383
384 default: workspace
385
386 - `BetaWorkspaceRateLimitOrganizationSource object`
387
388 - `type: "organization"`
389
390 Always `organization`: no workspace-level override is stored, so the organization's value applies.
391
392 default: organization
393
343394 - `value: number`
344395
345 The workspace-level override value for this limiter type.
396 The workspace's value for this limiter type: the workspace-level override when `source.type` is `workspace`, otherwise the organization's value.
346397
347398 - `models: array of string or null`
348399
from line 401
350401
351402 - `rate_limit_id: string`
352403
353 The `id` of the organization's RateLimit entry this override applies to.
404 The `id` of the organization's RateLimit entry this entry applies to.
354405
355406 - `workspace_id: string`
356407
357 ID of the Workspace this override applies to.
408 ID of the Workspace this entry applies to.
358409
359410 - `group_type: "batch" or "files" or "model_group" or 3 more`
360411
from line 425
374425
375426 - `"web_search"`
376427
428### Beta Workspace Rate Limit Organization Source
429
430- `BetaWorkspaceRateLimitOrganizationSource object`
431
432 - `type: "organization"`
433
434 Always `organization`: no workspace-level override is stored, so the organization's value applies.
435
436 default: organization
437
377438### Beta Workspace Rate Limit Value
378439
379440- `BetaWorkspaceRateLimitValue object`
from line 447
386447
387448 The organization-level value for the same limiter type, for reference. `null` when the organization has no limit configured for this limiter type.
388449
450 - `source: BetaWorkspaceRateLimitWorkspaceSource or BetaWorkspaceRateLimitOrganizationSource`
451
452 Where `value` comes from. `organization` values are listed only when `include_inherited` is `true`, and then `value` equals `org_limit`.
453
454 - `BetaWorkspaceRateLimitWorkspaceSource object`
455
456 - `type: "workspace"`
457
458 Always `workspace`: a workspace-level override is stored.
459
460 default: workspace
461
462 - `BetaWorkspaceRateLimitOrganizationSource object`
463
464 - `type: "organization"`
465
466 Always `organization`: no workspace-level override is stored, so the organization's value applies.
467
468 default: organization
469
389470 - `value: number`
390471
391 The workspace-level override value for this limiter type.
472 The workspace's value for this limiter type: the workspace-level override when `source.type` is `workspace`, otherwise the organization's value.
473
474### Beta Workspace Rate Limit Workspace Source
475
476- `BetaWorkspaceRateLimitWorkspaceSource object`
477
478 - `type: "workspace"`
479
480 Always `workspace`: a workspace-level override is stored.
481
482 default: workspace
392483
api/beta/organization/workspaces/rate_limits/list Changed · +41 / -10 lines
from line 7
77
88**GET** `/v1/organizations/workspaces/{workspace_id}/rate_limits`
99
10List rate-limit overrides configured for a workspace.
10List a workspace's rate limits.
1111
12Returns only the groups and limiter types that have a workspace-level
13override. Groups without overrides inherit the organization limits and
14are not listed; use `GET /v1/organizations/rate_limits` to see those.
12By default, returns only the groups and limiter types that have a
13workspace-level override. With `include_inherited=true`, returns every
14group with organization-level limits the workspace can see, listing for
15each the values it inherits from the organization as well as its own
16overrides. Each value's `source` says which it is.
1517
1618When `limit` is omitted, every matching entry is returned in a single
1719page; when `limit` truncates the result, follow `next_page` to fetch
from line 43
4143
4244 - `"web_search"`
4345
46- `include_inherited: optional boolean`
47
48 Also list the limiter values the workspace inherits from the organization, including groups with no workspace-level override.
49
50 default: false
51
4452- `limit: optional number`
4553
4654 Maximum number of items to return per page. Ranges from `1` to `1000`.
from line 65
5765
5866- `data: array of BetaWorkspaceRateLimit`
5967
60 Rate-limit entries for the workspace, one per group that has at least one override.
68 Rate-limit entries for the workspace: one per group with at least one override, or, with `include_inherited` set to `true`, one per group the workspace can see that has organization-level limits.
6169
6270 - `type: "workspace_rate_limit"`
6371
from line 155
147155
148156 - `limits: array of BetaWorkspaceRateLimitValue`
149157
150 The limiter values overridden for this group in this workspace. Limiter types without a workspace override are omitted and inherit the organization value.
158 The workspace's limiter values for this group. By default only the limiter types with a workspace-level override are listed. With `include_inherited` set to `true`, the limiter types the workspace inherits from the organization are listed too, each marked by `source`.
151159
152160 - `type: string`
153161
from line 165
157165
158166 The organization-level value for the same limiter type, for reference. `null` when the organization has no limit configured for this limiter type.
159167
168 - `source: BetaWorkspaceRateLimitWorkspaceSource or BetaWorkspaceRateLimitOrganizationSource`
169
170 Where `value` comes from. `organization` values are listed only when `include_inherited` is `true`, and then `value` equals `org_limit`.
171
172 - `BetaWorkspaceRateLimitWorkspaceSource object`
173
174 - `type: "workspace"`
175
176 Always `workspace`: a workspace-level override is stored.
177
178 default: workspace
179
180 - `BetaWorkspaceRateLimitOrganizationSource object`
181
182 - `type: "organization"`
183
184 Always `organization`: no workspace-level override is stored, so the organization's value applies.
185
186 default: organization
187
160188 - `value: number`
161189
162 The workspace-level override value for this limiter type.
190 The workspace's value for this limiter type: the workspace-level override when `source.type` is `workspace`, otherwise the organization's value.
163191
164192 - `models: array of string or null`
165193
from line 195
167195
168196 - `rate_limit_id: string`
169197
170 The `id` of the organization's RateLimit entry this override applies to.
198 The `id` of the organization's RateLimit entry this entry applies to.
171199
172200 - `workspace_id: string`
173201
174 ID of the Workspace this override applies to.
202 ID of the Workspace this entry applies to.
175203
176204 - `group_type: "batch" or "files" or "model_group" or 3 more`
177205
from line 246
218246 "limits": [
219247 {
220248 "org_limit": 0,
249 "source": {
250 "type": "workspace"
251 },
221252 "type": "type",
222253 "value": 0
223254 }
api/completions Changed · +8 / -6 lines
from line 17
1717
1818### Headers
1919
20- `"anthropic-workspace-id": optional string`
21
22 Optional header to select the Workspace for this request. The value is a Workspace ID (for example, `wrkspc_011CZkZaBF1tNoB5wlCeusgy`).
23
24 Only needed for credentials that can act on more than one Workspace. A credential that belongs to a specific Workspace may omit it; if sent, it must match that Workspace.
25
2026- `"anthropic-beta": optional array of AnthropicBeta`
2127
28 **Deprecated**: Deprecated. This parameter has no effect on this method and will be removed in a future release.
29
2230 Optional header to specify the beta version(s) you want to use.
2331
2432 - `string`
from line 128
120128 - `"inline-tools-2026-09-15"`
121129
122130 - `"mcp-client-2026-09-15"`
123
124- `"anthropic-workspace-id": optional string`
125
126 Optional header to select the Workspace for this request. The value is a Workspace ID (for example, `wrkspc_011CZkZaBF1tNoB5wlCeusgy`).
127
128 Only needed for credentials that can act on more than one Workspace. A credential that belongs to a specific Workspace may omit it; if sent, it must match that Workspace.
129131
130132### Body parameters
131133
api/completions/create Changed · +8 / -6 lines
from line 15
1515
1616## Headers
1717
18- `"anthropic-workspace-id": optional string`
19
20 Optional header to select the Workspace for this request. The value is a Workspace ID (for example, `wrkspc_011CZkZaBF1tNoB5wlCeusgy`).
21
22 Only needed for credentials that can act on more than one Workspace. A credential that belongs to a specific Workspace may omit it; if sent, it must match that Workspace.
23
1824- `"anthropic-beta": optional array of AnthropicBeta`
1925
26 **Deprecated**: Deprecated. This parameter has no effect on this method and will be removed in a future release.
27
2028 Optional header to specify the beta version(s) you want to use.
2129
2230 - `string`
from line 126
118126 - `"inline-tools-2026-09-15"`
119127
120128 - `"mcp-client-2026-09-15"`
121
122- `"anthropic-workspace-id": optional string`
123
124 Optional header to select the Workspace for this request. The value is a Workspace ID (for example, `wrkspc_011CZkZaBF1tNoB5wlCeusgy`).
125
126 Only needed for credentials that can act on more than one Workspace. A credential that belongs to a specific Workspace may omit it; if sent, it must match that Workspace.
127129
128130## Body parameters
129131
api/compliance Changed · +2545 / -1467 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 19
1919
2020#### Query parameters
2121
22- `activity_types: optional array of "abuse_decision_received" or "account_deleted" or "admin_api_key_created" or 507 more`
22- `activity_types: optional array of "abuse_decision_received" or "account_deleted" or "admin_api_key_created" or 511 more`
2323
2424 Filter activities by type. See the response `data` schema for the additional fields each type returns. Cannot be combined with `exclude_activity_types[]`.
2525
from line 251
251251
252252 An organization admin allowed one artifact to be shared outside the organization by link while the organization-wide external sharing setting was off, or revoked that permission.
253253
254 - `"claude_artifact_invite_accepted"`
255
256 Someone outside the organization signed in with a verified account for the invited address and accepted an invitation to an artifact, and can now open it; recorded in the artifact owner's organization. The person who accepted is identified by `invitee_email` and `invitee_user_id`.
257
258 - `"claude_artifact_invite_created"`
259
260 A member invited (or re-invited) an email address outside the organization to an artifact; recorded in the artifact owner's organization with the inviting member as the actor.
261
262 - `"claude_artifact_invite_revoked"`
263
264 A member withdrew an invitation to an artifact for someone outside the organization, removing any access that invitation had granted; recorded in the artifact owner's organization with that member as the actor.
265
266 - `"claude_artifact_invite_role_updated"`
267
268 A member changed the access level of an email invitation to an artifact for someone outside the organization, whether the invitation was still pending or had been accepted; recorded in the artifact owner's organization with that member as the actor.
269
254270 - `"claude_artifact_published"`
255271
256272 A new version of an artifact was published — for an artifact created in a chat this is the action that made it publicly viewable; for an artifact created outside a chat it is recorded when the artifact is saved, including saves of private artifacts, except that automatic saves made while a person keeps editing may be recorded periodically for that person rather than once per save; changes to who can access the artifact are recorded separately as claude_artifact_sharing_updated.
from line 1021
10051021
10061022 - `"mcp_server_managed_auth_token_exchanged"`
10071023
1008 A user attempted to obtain an access token for an MCP server via enterprise managed authorization. This event reports the outcomes of attempted token exchanges. Repeated failures with the same cause may be reported once until the cause changes, and requests denied by organization policy before a token exchange is attempted are not reported, with the exception of the "connector_scope_not_granted" failures described under error_type.
1024 A user attempted to obtain an access token for an MCP server via enterprise managed authorization. This event reports the outcomes of attempted token exchanges. Repeated failures with the same cause may be reported once until the cause changes, and requests refused before a token exchange is attempted are not reported, except the "connector_scope_not_granted" and "identity_assertion_refused" failures described under error_type.
10091025
10101026 - `"mcp_server_managed_auth_updated"`
10111027
from line 2134
21182134
21192135 format: date-time
21202136
2121- `exclude_activity_types: optional array of "abuse_decision_received" or "account_deleted" or "admin_api_key_created" or 507 more`
2137- `exclude_activity_types: optional array of "abuse_decision_received" or "account_deleted" or "admin_api_key_created" or 511 more`
21222138
21232139 Exclude activities of these types. Cannot be combined with `activity_types[]`.
21242140
from line 2366
23502366
23512367 An organization admin allowed one artifact to be shared outside the organization by link while the organization-wide external sharing setting was off, or revoked that permission.
23522368
2369 - `"claude_artifact_invite_accepted"`
2370
2371 Someone outside the organization signed in with a verified account for the invited address and accepted an invitation to an artifact, and can now open it; recorded in the artifact owner's organization. The person who accepted is identified by `invitee_email` and `invitee_user_id`.
2372
2373 - `"claude_artifact_invite_created"`
2374
2375 A member invited (or re-invited) an email address outside the organization to an artifact; recorded in the artifact owner's organization with the inviting member as the actor.
2376
2377 - `"claude_artifact_invite_revoked"`
2378
2379 A member withdrew an invitation to an artifact for someone outside the organization, removing any access that invitation had granted; recorded in the artifact owner's organization with that member as the actor.
2380
2381 - `"claude_artifact_invite_role_updated"`
2382
2383 A member changed the access level of an email invitation to an artifact for someone outside the organization, whether the invitation was still pending or had been accepted; recorded in the artifact owner's organization with that member as the actor.
2384
23532385 - `"claude_artifact_published"`
23542386
23552387 A new version of an artifact was published — for an artifact created in a chat this is the action that made it publicly viewable; for an artifact created outside a chat it is recorded when the artifact is saved, including saves of private artifacts, except that automatic saves made while a person keeps editing may be recorded periodically for that person rather than once per save; changes to who can access the artifact are recorded separately as claude_artifact_sharing_updated.
from line 3136
31043136
31053137 - `"mcp_server_managed_auth_token_exchanged"`
31063138
3107 A user attempted to obtain an access token for an MCP server via enterprise managed authorization. This event reports the outcomes of attempted token exchanges. Repeated failures with the same cause may be reported once until the cause changes, and requests denied by organization policy before a token exchange is attempted are not reported, with the exception of the "connector_scope_not_granted" failures described under error_type.
3139 A user attempted to obtain an access token for an MCP server via enterprise managed authorization. This event reports the outcomes of attempted token exchanges. Repeated failures with the same cause may be reported once until the cause changes, and requests refused before a token exchange is attempted are not reported, except the "connector_scope_not_granted" and "identity_assertion_refused" failures described under error_type.
31083140
31093141 - `"mcp_server_managed_auth_updated"`
31103142
from line 4241
42094241
42104242#### Returns
42114243
4212- `data: optional array of AbuseDecisionReceived or AccountDeleted or AdminAPIKeyCreated or 507 more`
4244- `data: optional array of AbuseDecisionReceived or AccountDeleted or AdminAPIKeyCreated or 511 more`
42134245
42144246 List of activity records. Each element's `type` field identifies which activity it is and which additional fields are present.
42154247
from line 7617
75857617
75867618 Unique identifier for the activity e.g. 'activity_abcd1234'
75877619
7620 - `actor_outside_organization: optional boolean or null`
7621
7622 True when the person who acted belongs to a different organization than the artifact's owner organization, such as someone invited by email who accepted commenter or editor access. Absent on older events; treat absence as false.
7623
75887624 - `claude_artifact_comment_id: optional string or null`
75897625
75907626 The comment's identifier. Present when the activity relates to a specific comment, for example a new comment, an author's edit of one, or an existing comment sent to Claude or withdrawn from Claude; absent for thread-level actions performed without a comment, such as resolve, reopen, deletion, an activation change, or a resolve by a Claude session.
from line 8589
85538589
85548590 Unique identifier for the activity e.g. 'activity_abcd1234'
85558591
8592 - `actor_outside_organization: optional boolean or null`
8593
8594 True when the person who published belongs to a different organization than the artifact's owner organization, such as someone invited by email who accepted editor access. Absent on older events; treat absence as false.
8595
85568596 - `claude_artifact_version_id: optional string or null`
85578597
85588598 The version identifier recorded as live by this publish.
from line 9769
97299769 gateway or a customer-registered federation issuer — acting without an
97309770 Anthropic-provisioned account or service account.
97319771
9732 - `type: optional "federated_actor"`
9733
9734 default: federated_actor
9735
9736 - `provider: FederatedActorAwsProvider or FederatedActorAzureProvider or FederatedActorGcpProvider or FederatedActorOidcProvider`
9737
9738 - `FederatedActorAwsProvider object`
9739
9740 Asserting party: the AWS account the organization is bound to.
9741
9742 - `type: optional "aws"`
9743
9744 default: aws
9745
9746 - `account_id: string`
9747
9748 - `signed_principal: string`
9749
9750 The AWS-signed ARN of the IAM principal that requested the token.
9751
9752 - `FederatedActorAzureProvider object`
9753
9754 Asserting party: the Azure subscription the organization is bound to.
9755
9756 - `type: optional "azure"`
9757
9758 default: azure
9759
9760 - `subscription_id: string`
9761
9762 - `FederatedActorGcpProvider object`
9763
9764 Asserting party: the GCP project the organization is bound to.
9765
9766 - `type: optional "gcp"`
9767
9768 default: gcp
9769
9770 - `project_number: string`
9771
9772 - `FederatedActorOidcProvider object`
9773
9774 Asserting party: a customer-registered OIDC federation issuer.
9775
9776 - `type: optional "oidc"`
9777
9778 default: oidc
9779
9780 - `issuer: optional string or null`
9781
9782 The federation issuer's URL. Null when the presented credential failed verification.
9783
9784 - `ip_address: optional string or null`
9785
9786 - `subject: optional string or null`
9787
9788 The provider's verified identifier for the caller; its form depends on the provider.
9789
9790 - `user_agent: optional string or null`
9791
9792 - `AttestedDeviceActor object`
9793
9794 An attested mobile device authenticated via Apple App Attest.
9795
9796 - `type: optional "attested_device_actor"`
9797
9798 default: attested_device_actor
9799
9800 - `external_client_id: string`
9801
9802 - `kid_hash: string`
9803
9804 - `ip_address: optional string or null`
9805
9806 - `user_agent: optional string or null`
9807
9808 - `id: optional string`
9809
9810 Unique identifier for the activity e.g. 'activity_abcd1234'
9811
9812 - `cc_email_count: optional number or null`
9813
9814 Number of 'cc' email recipients.
9815
9816 - `created_at: optional string`
9817
9818 When this activity occurred.
9819
9820 format: date-time
9821
9822 - `organization_id: optional string or null`
9823
9824 Organization ID this activity is associated with
9825
9826 - `organization_uuid: optional string or null`
9827
9828 Organization UUID where the activity occurred. Null when the activity is not tied to an organization (for example, login and logout events or calls to the Compliance API).
9829
9830 - `primary_email_set: optional boolean or null`
9831
9832 Whether a primary billing email is configured.
9833
9834 - `to_email_count: optional number or null`
9835
9836 Number of 'to' em
9772
api/compliance/activities Changed · +5018 / -2922 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 17
1717
1818### Query parameters
1919
20- `activity_types: optional array of "abuse_decision_received" or "account_deleted" or "admin_api_key_created" or 507 more`
20- `activity_types: optional array of "abuse_decision_received" or "account_deleted" or "admin_api_key_created" or 511 more`
2121
2222 Filter activities by type. See the response `data` schema for the additional fields each type returns. Cannot be combined with `exclude_activity_types[]`.
2323
from line 249
249249
250250 An organization admin allowed one artifact to be shared outside the organization by link while the organization-wide external sharing setting was off, or revoked that permission.
251251
252 - `"claude_artifact_invite_accepted"`
253
254 Someone outside the organization signed in with a verified account for the invited address and accepted an invitation to an artifact, and can now open it; recorded in the artifact owner's organization. The person who accepted is identified by `invitee_email` and `invitee_user_id`.
255
256 - `"claude_artifact_invite_created"`
257
258 A member invited (or re-invited) an email address outside the organization to an artifact; recorded in the artifact owner's organization with the inviting member as the actor.
259
260 - `"claude_artifact_invite_revoked"`
261
262 A member withdrew an invitation to an artifact for someone outside the organization, removing any access that invitation had granted; recorded in the artifact owner's organization with that member as the actor.
263
264 - `"claude_artifact_invite_role_updated"`
265
266 A member changed the access level of an email invitation to an artifact for someone outside the organization, whether the invitation was still pending or had been accepted; recorded in the artifact owner's organization with that member as the actor.
267
252268 - `"claude_artifact_published"`
253269
254270 A new version of an artifact was published — for an artifact created in a chat this is the action that made it publicly viewable; for an artifact created outside a chat it is recorded when the artifact is saved, including saves of private artifacts, except that automatic saves made while a person keeps editing may be recorded periodically for that person rather than once per save; changes to who can access the artifact are recorded separately as claude_artifact_sharing_updated.
from line 1019
10031019
10041020 - `"mcp_server_managed_auth_token_exchanged"`
10051021
1006 A user attempted to obtain an access token for an MCP server via enterprise managed authorization. This event reports the outcomes of attempted token exchanges. Repeated failures with the same cause may be reported once until the cause changes, and requests denied by organization policy before a token exchange is attempted are not reported, with the exception of the "connector_scope_not_granted" failures described under error_type.
1022 A user attempted to obtain an access token for an MCP server via enterprise managed authorization. This event reports the outcomes of attempted token exchanges. Repeated failures with the same cause may be reported once until the cause changes, and requests refused before a token exchange is attempted are not reported, except the "connector_scope_not_granted" and "identity_assertion_refused" failures described under error_type.
10071023
10081024 - `"mcp_server_managed_auth_updated"`
10091025
from line 2132
21162132
21172133 format: date-time
21182134
2119- `exclude_activity_types: optional array of "abuse_decision_received" or "account_deleted" or "admin_api_key_created" or 507 more`
2135- `exclude_activity_types: optional array of "abuse_decision_received" or "account_deleted" or "admin_api_key_created" or 511 more`
21202136
21212137 Exclude activities of these types. Cannot be combined with `activity_types[]`.
21222138
from line 2364
23482364
23492365 An organization admin allowed one artifact to be shared outside the organization by link while the organization-wide external sharing setting was off, or revoked that permission.
23502366
2367 - `"claude_artifact_invite_accepted"`
2368
2369 Someone outside the organization signed in with a verified account for the invited address and accepted an invitation to an artifact, and can now open it; recorded in the artifact owner's organization. The person who accepted is identified by `invitee_email` and `invitee_user_id`.
2370
2371 - `"claude_artifact_invite_created"`
2372
2373 A member invited (or re-invited) an email address outside the organization to an artifact; recorded in the artifact owner's organization with the inviting member as the actor.
2374
2375 - `"claude_artifact_invite_revoked"`
2376
2377 A member withdrew an invitation to an artifact for someone outside the organization, removing any access that invitation had granted; recorded in the artifact owner's organization with that member as the actor.
2378
2379 - `"claude_artifact_invite_role_updated"`
2380
2381 A member changed the access level of an email invitation to an artifact for someone outside the organization, whether the invitation was still pending or had been accepted; recorded in the artifact owner's organization with that member as the actor.
2382
23512383 - `"claude_artifact_published"`
23522384
23532385 A new version of an artifact was published — for an artifact created in a chat this is the action that made it publicly viewable; for an artifact created outside a chat it is recorded when the artifact is saved, including saves of private artifacts, except that automatic saves made while a person keeps editing may be recorded periodically for that person rather than once per save; changes to who can access the artifact are recorded separately as claude_artifact_sharing_updated.
from line 3134
31023134
31033135 - `"mcp_server_managed_auth_token_exchanged"`
31043136
3105 A user attempted to obtain an access token for an MCP server via enterprise managed authorization. This event reports the outcomes of attempted token exchanges. Repeated failures with the same cause may be reported once until the cause changes, and requests denied by organization policy before a token exchange is attempted are not reported, with the exception of the "connector_scope_not_granted" failures described under error_type.
3137 A user attempted to obtain an access token for an MCP server via enterprise managed authorization. This event reports the outcomes of attempted token exchanges. Repeated failures with the same cause may be reported once until the cause changes, and requests refused before a token exchange is attempted are not reported, except the "connector_scope_not_granted" and "identity_assertion_refused" failures described under error_type.
31063138
31073139 - `"mcp_server_managed_auth_updated"`
31083140
from line 4239
42074239
42084240### Returns
42094241
4210- `data: optional array of AbuseDecisionReceived or AccountDeleted or AdminAPIKeyCreated or 507 more`
4242- `data: optional array of AbuseDecisionReceived or AccountDeleted or AdminAPIKeyCreated or 511 more`
42114243
42124244 List of activity records. Each element's `type` field identifies which activity it is and which additional fields are present.
42134245
from line 7615
75837615
75847616 Unique identifier for the activity e.g. 'activity_abcd1234'
75857617
7618 - `actor_outside_organization: optional boolean or null`
7619
7620 True when the person who acted belongs to a different organization than the artifact's owner organization, such as someone invited by email who accepted commenter or editor access. Absent on older events; treat absence as false.
7621
75867622 - `claude_artifact_comment_id: optional string or null`
75877623
75887624 The comment's identifier. Present when the activity relates to a specific comment, for example a new comment, an author's edit of one, or an existing comment sent to Claude or withdrawn from Claude; absent for thread-level actions performed without a comment, such as resolve, reopen, deletion, an activation change, or a resolve by a Claude session.
from line 8587
85518587
85528588 Unique identifier for the activity e.g. 'activity_abcd1234'
85538589
8590 - `actor_outside_organization: optional boolean or null`
8591
8592 True when the person who published belongs to a different organization than the artifact's owner organization, such as someone invited by email who accepted editor access. Absent on older events; treat absence as false.
8593
85548594 - `claude_artifact_version_id: optional string or null`
85558595
85568596 The version identifier recorded as live by this publish.
from line 9767
97279767 gateway or a customer-registered federation issuer — acting without an
97289768 Anthropic-provisioned account or service account.
97299769
9730 - `type: optional "federated_actor"`
9731
9732 default: federated_actor
9733
9734 - `provider: FederatedActorAwsProvider or FederatedActorAzureProvider or FederatedActorGcpProvider or FederatedActorOidcProvider`
9735
9736 - `FederatedActorAwsProvider object`
9737
9738 Asserting party: the AWS account the organization is bound to.
9739
9740 - `type: optional "aws"`
9741
9742 default: aws
9743
9744 - `account_id: string`
9745
9746 - `signed_principal: string`
9747
9748 The AWS-signed ARN of the IAM principal that requested the token.
9749
9750 - `FederatedActorAzureProvider object`
9751
9752 Asserting party: the Azure subscription the organization is bound to.
9753
9754 - `type: optional "azure"`
9755
9756 default: azure
9757
9758 - `subscription_id: string`
9759
9760 - `FederatedActorGcpProvider object`
9761
9762 Asserting party: the GCP project the organization is bound to.
9763
9764 - `type: optional "gcp"`
9765
9766 default: gcp
9767
9768 - `project_number: string`
9769
9770 - `FederatedActorOidcProvider object`
9771
9772 Asserting party: a customer-registered OIDC federation issuer.
9773
9774 - `type: optional "oidc"`
9775
9776 default: oidc
9777
9778 - `issuer: optional string or null`
9779
9780 The federation issuer's URL. Null when the presented credential failed verification.
9781
9782 - `ip_address: optional string or null`
9783
9784 - `subject: optional string or null`
9785
9786 The provider's verified identifier for the caller; its form depends on the provider.
9787
9788 - `user_agent: optional string or null`
9789
9790 - `AttestedDeviceActor object`
9791
9792 An attested mobile device authenticated via Apple App Attest.
9793
9794 - `type: optional "attested_device_actor"`
9795
9796 default: attested_device_actor
9797
9798 - `external_client_id: string`
9799
9800 - `kid_hash: string`
9801
9802 - `ip_address: optional string or null`
9803
9804 - `user_agent: optional string or null`
9805
9806 - `id: optional string`
9807
9808 Unique identifier for the activity e.g. 'activity_abcd1234'
9809
9810 - `cc_email_count: optional number or null`
9811
9812 Number of 'cc' email recipients.
9813
9814 - `created_at: optional string`
9815
9816 When this activity occurred.
9817
9818 format: date-time
9819
9820 - `organization_id: optional string or null`
9821
9822 Organization ID this activity is associated with
9823
9824 - `organization_uuid: optional string or null`
9825
9826 Organization UUID where the activity occurred. Null when the activity is not tied to an organization (for example, login and logout events or calls to the Compliance API).
9827
9828 - `primary_email_set: optional boolean or null`
9829
9830 Whether a primary billing email is configured.
9831
9832 - `to_email_count: optional number or null`
9833
9834 Number of 'to' email recipients.
9835
9836 - `CcrAgentCreat
9770 - `type: optional "federated_actor
api/compliance/activities/list Changed · +2527 / -1463 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 15
1515
1616## Query parameters
1717
18- `activity_types: optional array of "abuse_decision_received" or "account_deleted" or "admin_api_key_created" or 507 more`
18- `activity_types: optional array of "abuse_decision_received" or "account_deleted" or "admin_api_key_created" or 511 more`
1919
2020 Filter activities by type. See the response `data` schema for the additional fields each type returns. Cannot be combined with `exclude_activity_types[]`.
2121
from line 247
247247
248248 An organization admin allowed one artifact to be shared outside the organization by link while the organization-wide external sharing setting was off, or revoked that permission.
249249
250 - `"claude_artifact_invite_accepted"`
251
252 Someone outside the organization signed in with a verified account for the invited address and accepted an invitation to an artifact, and can now open it; recorded in the artifact owner's organization. The person who accepted is identified by `invitee_email` and `invitee_user_id`.
253
254 - `"claude_artifact_invite_created"`
255
256 A member invited (or re-invited) an email address outside the organization to an artifact; recorded in the artifact owner's organization with the inviting member as the actor.
257
258 - `"claude_artifact_invite_revoked"`
259
260 A member withdrew an invitation to an artifact for someone outside the organization, removing any access that invitation had granted; recorded in the artifact owner's organization with that member as the actor.
261
262 - `"claude_artifact_invite_role_updated"`
263
264 A member changed the access level of an email invitation to an artifact for someone outside the organization, whether the invitation was still pending or had been accepted; recorded in the artifact owner's organization with that member as the actor.
265
250266 - `"claude_artifact_published"`
251267
252268 A new version of an artifact was published — for an artifact created in a chat this is the action that made it publicly viewable; for an artifact created outside a chat it is recorded when the artifact is saved, including saves of private artifacts, except that automatic saves made while a person keeps editing may be recorded periodically for that person rather than once per save; changes to who can access the artifact are recorded separately as claude_artifact_sharing_updated.
from line 1017
10011017
10021018 - `"mcp_server_managed_auth_token_exchanged"`
10031019
1004 A user attempted to obtain an access token for an MCP server via enterprise managed authorization. This event reports the outcomes of attempted token exchanges. Repeated failures with the same cause may be reported once until the cause changes, and requests denied by organization policy before a token exchange is attempted are not reported, with the exception of the "connector_scope_not_granted" failures described under error_type.
1020 A user attempted to obtain an access token for an MCP server via enterprise managed authorization. This event reports the outcomes of attempted token exchanges. Repeated failures with the same cause may be reported once until the cause changes, and requests refused before a token exchange is attempted are not reported, except the "connector_scope_not_granted" and "identity_assertion_refused" failures described under error_type.
10051021
10061022 - `"mcp_server_managed_auth_updated"`
10071023
from line 2130
21142130
21152131 format: date-time
21162132
2117- `exclude_activity_types: optional array of "abuse_decision_received" or "account_deleted" or "admin_api_key_created" or 507 more`
2133- `exclude_activity_types: optional array of "abuse_decision_received" or "account_deleted" or "admin_api_key_created" or 511 more`
21182134
21192135 Exclude activities of these types. Cannot be combined with `activity_types[]`.
21202136
from line 2362
23462362
23472363 An organization admin allowed one artifact to be shared outside the organization by link while the organization-wide external sharing setting was off, or revoked that permission.
23482364
2365 - `"claude_artifact_invite_accepted"`
2366
2367 Someone outside the organization signed in with a verified account for the invited address and accepted an invitation to an artifact, and can now open it; recorded in the artifact owner's organization. The person who accepted is identified by `invitee_email` and `invitee_user_id`.
2368
2369 - `"claude_artifact_invite_created"`
2370
2371 A member invited (or re-invited) an email address outside the organization to an artifact; recorded in the artifact owner's organization with the inviting member as the actor.
2372
2373 - `"claude_artifact_invite_revoked"`
2374
2375 A member withdrew an invitation to an artifact for someone outside the organization, removing any access that invitation had granted; recorded in the artifact owner's organization with that member as the actor.
2376
2377 - `"claude_artifact_invite_role_updated"`
2378
2379 A member changed the access level of an email invitation to an artifact for someone outside the organization, whether the invitation was still pending or had been accepted; recorded in the artifact owner's organization with that member as the actor.
2380
23492381 - `"claude_artifact_published"`
23502382
23512383 A new version of an artifact was published — for an artifact created in a chat this is the action that made it publicly viewable; for an artifact created outside a chat it is recorded when the artifact is saved, including saves of private artifacts, except that automatic saves made while a person keeps editing may be recorded periodically for that person rather than once per save; changes to who can access the artifact are recorded separately as claude_artifact_sharing_updated.
from line 3132
31003132
31013133 - `"mcp_server_managed_auth_token_exchanged"`
31023134
3103 A user attempted to obtain an access token for an MCP server via enterprise managed authorization. This event reports the outcomes of attempted token exchanges. Repeated failures with the same cause may be reported once until the cause changes, and requests denied by organization policy before a token exchange is attempted are not reported, with the exception of the "connector_scope_not_granted" failures described under error_type.
3135 A user attempted to obtain an access token for an MCP server via enterprise managed authorization. This event reports the outcomes of attempted token exchanges. Repeated failures with the same cause may be reported once until the cause changes, and requests refused before a token exchange is attempted are not reported, except the "connector_scope_not_granted" and "identity_assertion_refused" failures described under error_type.
31043136
31053137 - `"mcp_server_managed_auth_updated"`
31063138
from line 4237
42054237
42064238## Returns
42074239
4208- `data: optional array of AbuseDecisionReceived or AccountDeleted or AdminAPIKeyCreated or 507 more`
4240- `data: optional array of AbuseDecisionReceived or AccountDeleted or AdminAPIKeyCreated or 511 more`
42094241
42104242 List of activity records. Each element's `type` field identifies which activity it is and which additional fields are present.
42114243
from line 7613
75817613
75827614 Unique identifier for the activity e.g. 'activity_abcd1234'
75837615
7616 - `actor_outside_organization: optional boolean or null`
7617
7618 True when the person who acted belongs to a different organization than the artifact's owner organization, such as someone invited by email who accepted commenter or editor access. Absent on older events; treat absence as false.
7619
75847620 - `claude_artifact_comment_id: optional string or null`
75857621
75867622 The comment's identifier. Present when the activity relates to a specific comment, for example a new comment, an author's edit of one, or an existing comment sent to Claude or withdrawn from Claude; absent for thread-level actions performed without a comment, such as resolve, reopen, deletion, an activation change, or a resolve by a Claude session.
from line 8585
85498585
85508586 Unique identifier for the activity e.g. 'activity_abcd1234'
85518587
8588 - `actor_outside_organization: optional boolean or null`
8589
8590 True when the person who published belongs to a different organization than the artifact's owner organization, such as someone invited by email who accepted editor access. Absent on older events; treat absence as false.
8591
85528592 - `claude_artifact_version_id: optional string or null`
85538593
85548594 The version identifier recorded as live by this publish.
from line 9765
97259765 gateway or a customer-registered federation issuer — acting without an
97269766 Anthropic-provisioned account or service account.
97279767
9728 - `type: optional "federated_actor"`
9729
9730 default: federated_actor
9731
9732 - `provider: FederatedActorAwsProvider or FederatedActorAzureProvider or FederatedActorGcpProvider or FederatedActorOidcProvider`
9733
9734 - `FederatedActorAwsProvider object`
9735
9736 Asserting party: the AWS account the organization is bound to.
9737
9738 - `type: optional "aws"`
9739
9740 default: aws
9741
9742 - `account_id: string`
9743
9744 - `signed_principal: string`
9745
9746 The AWS-signed ARN of the IAM principal that requested the token.
9747
9748 - `FederatedActorAzureProvider object`
9749
9750 Asserting party: the Azure subscription the organization is bound to.
9751
9752 - `type: optional "azure"`
9753
9754 default: azure
9755
9756 - `subscription_id: string`
9757
9758 - `FederatedActorGcpProvider object`
9759
9760 Asserting party: the GCP project the organization is bound to.
9761
9762 - `type: optional "gcp"`
9763
9764 default: gcp
9765
9766 - `project_number: string`
9767
9768 - `FederatedActorOidcProvider object`
9769
9770 Asserting party: a customer-registered OIDC federation issuer.
9771
9772 - `type: optional "oidc"`
9773
9774 default: oidc
9775
9776 - `issuer: optional string or null`
9777
9778 The federation issuer's URL. Null when the presented credential failed verification.
9779
9780 - `ip_address: optional string or null`
9781
9782 - `subject: optional string or null`
9783
9784 The provider's verified identifier for the caller; its form depends on the provider.
9785
9786 - `user_agent: optional string or null`
9787
9788 - `AttestedDeviceActor object`
9789
9790 An attested mobile device authenticated via Apple App Attest.
9791
9792 - `type: optional "attested_device_actor"`
9793
9794 default: attested_device_actor
9795
9796 - `external_client_id: string`
9797
9798 - `kid_hash: string`
9799
9800 - `ip_address: optional string or null`
9801
9802 - `user_agent: optional string or null`
9803
9804 - `id: optional string`
9805
9806 Unique identifier for the activity e.g. 'activity_abcd1234'
9807
9808 - `cc_email_count: optional number or null`
9809
9810 Number of 'cc' email recipients.
9811
9812 - `created_at: optional string`
9813
9814 When this activity occurred.
9815
9816 format: date-time
9817
9818 - `organization_id: optional string or null`
9819
9820 Organization ID this activity is associated with
9821
9822 - `organization_uuid: optional string or null`
9823
9824 Organization UUID where the activity occurred. Null when the activity is not tied to an organization (for example, login and logout events or calls to the Compliance API).
9825
9826 - `primary_email_set: optional boolean or null`
9827
9828 Whether a primary billing email is configured.
9829
9830 - `to_email_count: optional number or null`
9831
9832 Number of 'to' email recipients.
9833
9834 - `CcrAgentC
9768 - `type: optional "federated_a
api/compliance/apps Changed · +15 / -3 lines
from line 15
1515compliance review. Results are sorted chronologically (time ascending)
1616by the `order_by` key, with ties broken by id.
1717
18Incremental polling with `order_by=updated_at` returns a chat again
19after it receives a new message, is moved into or out of a project, or
20is deleted in claude.ai. A chat is not guaranteed to be returned again
21after other edits, such as a rename.
22
1823**Deprecation notice:** Combining `user_ids[]` with any `updated_at.*`
1924filter is deprecated and will be rejected with HTTP 400 after
20252026-09-22. For incremental polling by update time, omit `user_ids[]`
from line 169
164169
165170 - `updated_at: string`
166171
167 Last update timestamp
172 Last update timestamp. Updated when the chat receives a new message, is moved into or out of a project, or is deleted in claude.ai. Other edits, such as renaming the chat, are not guaranteed to change it.
168173
169174 format: date-time
170175
from line 665
660665
661666- `updated_at: string`
662667
663 Last update timestamp
668 Last update timestamp. Updated when the chat receives a new message, is moved into or out of a project, or is deleted in claude.ai. Other edits, such as renaming the chat, are not guaranteed to change it.
664669
665670 format: date-time
666671
from line 2326
23212326honored for 24 hours: a cursor older than that is rejected with an
23222327explicit 400; restart the walk to read under the current boundary.
23232328
2329On a very large session, some pages are too large to read and return a
2330400; retrying does not help. If the request used `order=desc`, read the
2331session oldest first from its first page instead (omit `order` and
2332`page`, then follow `next_page`). Rarely, an oldest-first page returns
2333this 400 too; contact Anthropic support and quote the `request-id`
2334response header.
2335
23242336#### Path parameters
23252337
23262338- `local_session_id: string`
from line 2347
23352347
23362348- `order: optional "asc" or "desc"`
23372349
2338 Sort direction. `asc` (oldest-first, default) or `desc`.
2350 Sort direction. `asc` (oldest-first, default) or `desc`. On very large sessions some pages are too large to read and return a 400, far more often with `desc`; read those sessions with `asc`, starting again from the first page.
23392351
23402352 default: asc
23412353
api/compliance/apps/chats Changed · +8 / -3 lines
from line 13
1313compliance review. Results are sorted chronologically (time ascending)
1414by the `order_by` key, with ties broken by id.
1515
16Incremental polling with `order_by=updated_at` returns a chat again
17after it receives a new message, is moved into or out of a project, or
18is deleted in claude.ai. A chat is not guaranteed to be returned again
19after other edits, such as a rename.
20
1621**Deprecation notice:** Combining `user_ids[]` with any `updated_at.*`
1722filter is deprecated and will be rejected with HTTP 400 after
18232026-09-22. For incremental polling by update time, omit `user_ids[]`
from line 167
162167
163168 - `updated_at: string`
164169
165 Last update timestamp
170 Last update timestamp. Updated when the chat receives a new message, is moved into or out of a project, or is deleted in claude.ai. Other edits, such as renaming the chat, are not guaranteed to change it.
166171
167172 format: date-time
168173
from line 343
338343
339344 - `updated_at: string`
340345
341 Last update timestamp
346 Last update timestamp. Updated when the chat receives a new message, is moved into or out of a project, or is deleted in claude.ai. Other edits, such as renaming the chat, are not guaranteed to change it.
342347
343348 format: date-time
344349
from line 761
756761
757762- `updated_at: string`
758763
759 Last update timestamp
764 Last update timestamp. Updated when the chat receives a new message, is moved into or out of a project, or is deleted in claude.ai. Other edits, such as renaming the chat, are not guaranteed to change it.
760765
761766 format: date-time
762767
api/compliance/apps/chats/list Changed · +6 / -1 lines
from line 11
1111compliance review. Results are sorted chronologically (time ascending)
1212by the `order_by` key, with ties broken by id.
1313
14Incremental polling with `order_by=updated_at` returns a chat again
15after it receives a new message, is moved into or out of a project, or
16is deleted in claude.ai. A chat is not guaranteed to be returned again
17after other edits, such as a rename.
18
1419**Deprecation notice:** Combining `user_ids[]` with any `updated_at.*`
1520filter is deprecated and will be rejected with HTTP 400 after
16212026-09-22. For incremental polling by update time, omit `user_ids[]`
from line 165
160165
161166 - `updated_at: string`
162167
163 Last update timestamp
168 Last update timestamp. Updated when the chat receives a new message, is moved into or out of a project, or is deleted in claude.ai. Other edits, such as renaming the chat, are not guaranteed to change it.
164169
165170 format: date-time
166171
api/compliance/apps/sessions Changed · +8 / -1 lines
from line 257
257257honored for 24 hours: a cursor older than that is rejected with an
258258explicit 400; restart the walk to read under the current boundary.
259259
260On a very large session, some pages are too large to read and return a
261400; retrying does not help. If the request used `order=desc`, read the
262session oldest first from its first page instead (omit `order` and
263`page`, then follow `next_page`). Rarely, an oldest-first page returns
264this 400 too; contact Anthropic support and quote the `request-id`
265response header.
266
260267#### Path parameters
261268
262269- `local_session_id: string`
from line 278
271278
272279- `order: optional "asc" or "desc"`
273280
274 Sort direction. `asc` (oldest-first, default) or `desc`.
281 Sort direction. `asc` (oldest-first, default) or `desc`. On very large sessions some pages are too large to read and return a 400, far more often with `desc`; read those sessions with `asc`, starting again from the first page.
275282
276283 default: asc
277284
api/compliance/apps/sessions/local Changed · +8 / -1 lines
from line 371
371371honored for 24 hours: a cursor older than that is rejected with an
372372explicit 400; restart the walk to read under the current boundary.
373373
374On a very large session, some pages are too large to read and return a
375400; retrying does not help. If the request used `order=desc`, read the
376session oldest first from its first page instead (omit `order` and
377`page`, then follow `next_page`). Rarely, an oldest-first page returns
378this 400 too; contact Anthropic support and quote the `request-id`
379response header.
380
374381#### Path parameters
375382
376383- `local_session_id: string`
from line 392
385392
386393- `order: optional "asc" or "desc"`
387394
388 Sort direction. `asc` (oldest-first, default) or `desc`.
395 Sort direction. `asc` (oldest-first, default) or `desc`. On very large sessions some pages are too large to read and return a 400, far more often with `desc`; read those sessions with `asc`, starting again from the first page.
389396
390397 default: asc
391398