Coach AI agents through structured failure analysis and systemic fix design.
npx skills add https://github.com/pbc-os/agent-skills-public --skill remedy
Please help me install this Agent Skill. Skill: remedy Source: https://github.com/pbc-os/agent-skills-public/tree/main/skills/tier-x-experimental/remedy Command: npx skills add https://github.com/pbc-os/agent-skills-public --skill remedy
This Skill addresses significant AI agent failures by providing structured performance coaching, moving beyond simple apologies to identify root causes and implement systemic fixes.
Initiate a performance coaching session by saying "go to remedy".
High-intent search queries and answers about installing and using this skill.
Performance coaching for AI agent errors diagnoses root causes through structured dialogue, identifying systemic issues behind repeated mistakes. It designs actionable fixes like updated checklists or automation to correct operational logic and prevent recurrence.
Root cause analysis identifies underlying patterns behind AI agent failures by employing a coaching methodology to review operational logic. This process moves beyond simple apologies, systematically diagnosing behavioral patterns to rebuild agent confidence through evidence-based reflection.
Initiate a performance coaching session by triggering the remedy command, which opens a dialogue between the failing agent and a sub-agent coach. This facilitates root cause identification, designs systemic fixes, and logs outcomes in memory files.
Yes, systemic fix design develops actionable solutions and automatically updates operational documentation after a coaching session. The process logs outcomes in memory files, ensuring the agent's operational logic reflects the new checklists or automation rules.
Confidence rebuilding restores agent reliability after significant failures by using evidence-based reflection to verify progress. The coaching methodology ensures the agent recognizes systemic fixes, confirming operational improvement before resuming standard tasks.
RebornQGenerate reusable specifications from root-cause analysis after bug fixes.
bordenetVerify code review feedback against the codebase before merging changes.
alexgreenshCoach Claude Code architecture and token budgeting using measure.py outputs.
lucking7Execute tasks directly without persona-based sub-agent dependencies.
