Add a generalized stop-gate rule and synchronize it across autonomy rules and classifier prompt.
npx skills add https://github.com/npow/claude-skills --skill update-stop-gate
Please help me install this Agent Skill. Skill: update-stop-gate Source: https://github.com/npow/claude-skills/tree/main/update-stop-gate Command: npx skills add https://github.com/npow/claude-skills --skill update-stop-gate
This Skill prevents autonomy rules from drifting out of sync with the classifier that detects stalled or unsafe agent behavior, ensuring new stop-gate rules are generalized, consistently implemented, and syntax-validated.
Ask the update-stop-gate skill to add a stop-gate rule for the described agent behavior and verify both synchronized files.
High-intent search queries and answers about installing and using this skill.
To keep autonomy rules and classifier prompts in sync, update both the human-readable rules and the mechanically evaluable Python classifier checks together. This prevents drift when adding new intervention triggers for unsafe agent behavior.
A stop-gate rule is a generalized intervention condition that converts a specific bad-behavior incident into a reusable behavior class. It detects stalled workflows or inadequate fixes and triggers an intervention, while allowing narrow legitimate exceptions.
To add stall detection rules safely, update the Python classifier check and run a compile-check after changes. Ensure the updated hook passes Python syntax validation and preserves OSS-safe language by scrubbing proprietary wording.
When an agent repeatedly masks symptoms instead of fixing root causes, it indicates a missing stop-gate rule. You need to add a generalized behavior class and corresponding classifier check to trigger an intervention for this specific pattern.
Yes, you can fix both the autonomy rules and the classifier prompt with a single synchronized update. This approach applies matching guidance to the human-readable rules and mechanically evaluable checks to the Python hook simultaneously.
cslsoftwaresDefine and enforce AI agent autonomy policies for Providers.2.1.0.
shahmirerrorRegister and orchestrate named validation commands for CI-style code-quality checks.
krzemienskiAutomate validation gate execution with expo-mcp visual verification and hard-stop enforcement.
MovchanetsRefactor tangled code into cleaner implementations with incremental regression checks.
