Repository navigation
fix: mount new inputs before replaying programmatic calls - #371
Draft
JamieStivala wants to merge 1 commit into
Draft
JamieStivala wants to merge 1 commit into
JamieStivala wants to merge 1 commit into
Conversation
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Problem and change
A programmatic script may pause for an MCP download after execution has started. The newly uploaded file is not in the initial input set, and current continuations carry only tool results, so the script cannot open the download after the tool returns.
Pending replay responses now advertise
continuation_files: 1. Each issued tool result can include owned input references using the existing reference schema. Under the execution lock, the API validates ownership and limits, rejects filename replacement, and commits the additive file set together with immutable history before replay. The next sandbox iteration mounts the inputs before resuming user code. Identical retries remain idempotent. Blocking mode rejects the unsupported capability explicitly.Fixes #370. Related host report: LibreChat-AI/LibreChat#17092. Coordinated SDK counterpart: LibreChat-AI/agents#645. The focused branch excludes JA Docker packaging, registry scopes, and CI/CD changes.
Verification
The two narrow typing fixes outside replay preserve existing behavior and allow strict typechecking of legacy-field compatibility code.
References: MCP embedded tool resources, the continuation contract added in
docs/MCP_CONTINUATION_FILES.md, and the repository's existing file-reference authorization boundary.This is my first contribution to this repository, so any feedback is appreciated :) AI assistance was used; reproduction and verification limits are documented above.