fix: add anti-imitation instruction to recall tool description#411
Merged
Conversation
The model sometimes writes text mimicking recall markers (e.g. '📚 Fetching details for t:... simultaneously…') instead of calling the recall tool. Add an explicit instruction to RECALL_TOOL_DESCRIPTION telling the model never to write these markers itself. Closes #409
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Prevents the model from writing text that mimics recall markers instead of actually calling the recall tool.
Problem
In CM-1 eval (live mode), the model sometimes writes text like
📚 Fetching details for t:e8a949d7... and t:36d96080... simultaneously…as its final answer instead of invoking the recall tool. The model sees recall markers in prior assistant turns and imitates the pattern.Changes
RECALL_TOOL_DESCRIPTIONinpackages/core/src/recall.ts, telling the model never to write recall status markers itself — they are injected by the system automaticallyCloses #409