Audits skill descriptions to align triggers with intended actions and recompiles CLIs via LiteCLI.
npx skills add https://github.com/ahostbr/dotcodex --skill self-improve-ahostbr
Please help me install this Agent Skill. Skill: self-improve Source: https://github.com/ahostbr/dotcodex/tree/main/skills/self-improve Command: npx skills add https://github.com/ahostbr/dotcodex --skill self-improve-ahostbr
Orchestrates the self-improvement meta loop: scan all skills, identify underperformers, train their descriptions via /train, and recompile via LiteCLI so the agent's own interface gets sharper every cycle. Triggers on 'self-improve', 'improve my skills', 'sharpen skills', 'optimize descriptions', 'train all skills', 'meta improve', 'self improve loop', 'improve skill descriptions', 'run self-improvement', 'make skills better'.
Trigger the self-improvement loop to scan all skills, train underperformers, and recompile affected CLIs.
High-intent search queries and answers about installing and using this skill.
To optimize skill descriptions automatically, trigger the self-improvement loop to scan all skills, identify underperformers via an eval framework, and train their descriptions to align triggers with intended actions.
The self-improvement meta loop is an automated cycle that audits skill descriptions, evaluates trigger accuracy, prioritizes underperformers, trains them via /train, and recompiles affected CLIs through LiteCLI for sharper agent interfaces.
Yes, you need LiteCLI to recompile CLIs after training skill descriptions. The self-improvement loop requires access to LiteCLI for recompilation to ensure the agent's own interface updates when descriptions change.
Evaluating trigger accuracy scales across all skills by scanning the local catalog, collecting current descriptions, measuring accuracy via an eval framework, and prioritizing the worst performers for training to systematically improve the entire skill set.
Yes, you can prioritize underperforming skills for training. The self-improvement loop audits all skills, measures trigger accuracy, and specifically targets underperformers via /train before recompiling affected CLIs to report improvements.
Limitations of the self-improvement loop include requiring an eval framework to measure trigger accuracy and LiteCLI access for recompilation. Without these dependencies, the loop cannot audit descriptions, train underperformers, or recompile affected CLIs.
ahostbrScans installed skills, evaluates trigger accuracy, and retrains underperformers via /train.
brucebanner010198-commitsRun structured evaluations measuring trigger accuracy, stability, and gate behavior for Claude Skills.
HerxinsasaAutomate AI skill creation, evaluation, and iterative improvement workflows.
jcdendriteAudit Claude Code SKILL.md files for trigger accuracy and frontmatter consistency.
