fix(instructions): prevent model from echoing injected banner#603
Open
Tanmay9223 wants to merge 2 commits into
Open
fix(instructions): prevent model from echoing injected banner#603Tanmay9223 wants to merge 2 commits into
Tanmay9223 wants to merge 2 commits into
Conversation
…etrichGebert#438) Update the skills/ponytail-help/SKILL.md documentation which incorrectly stated that OpenCode ships 2 commands. It now accurately states that OpenCode ships all 6 commands. Re-run scripts/build-openclaw-skills.js to sync the updated documentation to the generated .opencode/command/ponytail-help.md file so tests pass.
…ietrichGebert#595) Add a "no self-reference" rule to the canonical AGENTS.md, the ponytail SKILL.md, and the fallback instructions. This prevents the model from hallucinating fake hook output when no conversational prompt is given. - Add the rule to skills/ponytail/SKILL.md - Add the rule to hooks/ponytail-instructions.js - Update AGENTS.md and all per-platform rule copies - Add invariant check to scripts/check-rule-copies.js
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
🎯 What: Adds a "no self-reference" rule to the canonical
AGENTS.md,skills/ponytail/SKILL.md, and fallback instructions.💡 Why: To prevent the model from echoing the injected ponytail banner as fake hook output (e.g. inventing a
_ctx_hookor<system-reminder>) when it receives an instruction-only prompt.✅ Verification: Re-synced all rule copies using
scripts/check-rule-copies.js, added a check for the new rule invariant, and verified all tests pass (npm test).✨ Result: The model will no longer hallucinate fake hook logs or echo the instructions back when invoked via a skill or on an empty turn.
Closes #595