Detect skill gaps and propose improvements across user, project, and repo scopes.
npx skills add https://github.com/fredrick84823/fstack --skill improve-fredrick84823
Please help me install this Agent Skill. Skill: improve Source: https://github.com/fredrick84823/fstack/tree/main/skills/improve Command: npx skills add https://github.com/fredrick84823/fstack --skill improve-fredrick84823
Automates continual learning and self-improvement of skills during use by detecting gaps and prompting for improvements.
Run /improve to start the self-improvement workflow for your skills.
High-intent search queries and answers about installing and using this skill.
Automating self-improvement for Claude skills requires detecting gaps during use and attributing them to downstream rules. This skill runs a 7-step workflow to automatically capture signals, build candidate proposals, and enforce human review gates before writing changes to SKILL.md.
Signal collection for skill management captures improvement signals from user interactions and skill outputs. It centralizes pending gaps into signal-queue.md and candidate-queue.md files, automatically detecting areas needing refinement across user, project, and repo scopes.
Claude hook integration supports automatically proposing skill improvements by leveraging Stop and SessionStart hooks. These automation hooks capture signals during active sessions and generate candidate proposals for your review without manual triggering.
Running self-improving workflows requires the SKILL.md file structure and supports scripts, references, and assets directories. You also need the jq dependency installed to process the JSON data used in signal queues and scope resolution.
Managing skill gaps across different project scopes requires scope resolution to localize changes and prevent cross-repo interference. This skill determines whether a gap belongs to the user, project, or repo scope, then routes the proposed improvement to the correct level.
A self-improving skill workflow requires a human review gate to enforce strict evaluation before any changes are applied. The system builds candidate proposals in memory and only writes back to SKILL.md after explicit human approval and changelog updates.
jaypatrickDiscover and install agent skills using Skills CLI commands.
FerroxLabsCheck for Ferrox updates and install a selected release channel after previewing changelogs.
theofernandezzCapture runtime signals and generate structured improvement entries for AI library maintenance.
sahinDiagnose GTM performance gaps from cross-skill data outputs.
