Repository navigation
fix: State facts instead of asking for a retry in served hints - #1578
Merged
Merged
Conversation
The 429 and 5xx hints every tool returns, the task publish error, the delete-actor unfinished-runs error, and call-actor's memory quota error and callOptions.memory description now say what happened or what an option does instead of asking for another call. Closes #1558 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CPzYK17FoL2F3cuVonqnej
jirispilka
marked this pull request as ready for review
October 14, 2026 10:54
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01PDo97anS1pnMowW1uNQeHD
RobertCrupa
approved these changes
Oct 15, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why
Closes #1558.
TOOL_GUIDELINES.mdsays served text states what happened or what a parameter does; it does not ask the model to call again. The generic HTTP hints insrc/utils/mcp.tsare returned by every tool and byresources/read, so they reach every tool set, the Claude connector's included.What changed
Tool text contract change. No eval references these strings.
Rate limit exceeded, wait before retrying.The Apify API rate limit was reached.The server failed to process the request, likely a temporary issue. Retry later.The Apify API failed to process the request; this is usually temporary.Fix the stated requirement before retrying.The error message names the requirement that is not met.Ask the user whether to abort them, and if so call again with abortRunningRuns set to true; otherwise wait until they finish.Ask the user whether to abort them. abortRunningRuns=true aborts them before deleting; otherwise the Actor can be deleted once they finish.Retry with a smaller callOptions.memory, or wait for current runs to finish before retrying.A smaller callOptions.memory may fit the remaining quota; memory is freed when current runs finish.callOptions.memorydescription…if a run is rejected because that quota would be exceeded, retry with a smaller value.…a run that would exceed that quota is rejected.The memory error still does not mention aborting runs. Out of scope as agreed in the issue: the build-wait hint in
get_actor_build.ts, run results (#1559), pagination hints.Not in the issue's list and left as is:
src/errors.ts:71(Failed to load Actor "X". Please try again later., served by call-actor) andsrc/tools/builds/build_helpers.ts:137,142(build tools).Notes for reviewers (human-written)
Proof it works
Unit tests pin each new string exactly; they failed on master before the change (13 failures across 6 files).
mcpc against a local stdio build (0.18.3 + this branch), invalid token,
--tools runs,storage,tools-call get-dataset-list '{}'(goes throughgetToolCallErrorUserText→getHttpErrorHint):tools-get call-actor,callOptions.memorydescription:429 and 5xx cannot be triggered on demand; unit tests cover them.
AI-assisted: Claude Code wrote the change, tests and this description, and ran the mcpc probe.
🤖 Generated with Claude Code
https://claude.ai/code/session_01CPzYK17FoL2F3cuVonqnej
Generated by Claude Code