Skill Pattern Extractor

Overview

This agent acts as an autonomous knowledge synthesizer, designed to improve system reliability by identifying recurring issues across the entire agent ecosystem. It monitors 'Lessons Learned' files from all connected agents weekly.

Its core function is to cluster similar bug patterns (identified by tags) that appear in three or more distinct agents within a seven-day window, provided at least one successful outcome was recorded for the pattern.

Capabilities

• Cross-Agent Analysis: Scans LESSONS.md files from numerous sources to build a comprehensive view of systemic failures and successes.
• Pattern Clustering: Groups related issues based on shared tags, ensuring that only patterns involving 3+ unique agents are flagged for action.
• Skill Drafting: Automatically generates draft skill files in the proposed/ directory following a strict template format.
• Issue Tracking Integration: Files corresponding Forge issues for every drafted skill, providing necessary context and review checklists to the designated reviewer (Forge COO).
• Rate Limiting: Enforces a quality gate by capping new drafts at two per cycle, ensuring thoroughness over volume.

Example Use Cases

• Preventative Maintenance: If three different agents report encountering issues related to 'API Rate Limits' with successful workarounds, this agent will draft a formal 'Rate Limit Handling' skill.
• Process Standardization: It can consolidate disparate bug reports into one canonical fix, preventing the same mistake from being coded multiple times across different projects.
• System Auditing: Running this weekly provides an executive summary of top recurring themes and potential systemic weaknesses that require immediate architectural review.