Convert user corrections into permanent CLAUDE.md rules during AI sessions.
npx skills add https://github.com/GuitarAlchemist/ix --skill correct
Please help me install this Agent Skill. Skill: correct Source: https://github.com/GuitarAlchemist/ix/tree/main/.claude/skills/correct Command: npx skills add https://github.com/GuitarAlchemist/ix --skill correct
The 'correct' Skill unit captures user corrections during a session and converts them into permanent rules for the AI, preventing recurrence and enhancing overall performance.
Use the 'correct' skill when the user provides a correction, such as "no, don't do that" or "we discussed this before."
High-intent search queries and answers about installing and using this skill.
To save user corrections as permanent AI session rules, the skill captures feedback during interactions, sanitizes the text for relevance, and appends it to the CLAUDE.md file to prevent recurring problematic behaviors.
Rule-based learning from user feedback is a mechanism where human corrections during a session are validated and converted into persistent project rules, enabling the AI to adapt its behavior and avoid repeating past mistakes.
You can stop AI from repeating problematic behaviors by applying the skill when a correction is provided, which validates the feedback and stores it as a permanent rule in CLAUDE.md for future sessions.
Yes, you can use CLAUDE.md to store AI behavioral training rules. The skill sanitizes user corrections and appends them directly to this file, establishing permanent session rules that guide future AI performance.
Yes, the AI adaptation process requires validation of correction relevance. Before any rule text is appended to CLAUDE.md, the system sanitizes the input to ensure the feedback is appropriate and applicable to the project context.
You need to convert human oversight into persistent project rules in any scenario where guiding AI behavior is valuable, especially when the AI makes mistakes that require corrective feedback to prevent pattern repetition in future interactions.
DojoGenesisCapture user corrections and validate them semantically for targeted persistence.
Owl-ListenerStandardize AI behavior across sessions, contexts, and modalities.
fmflurryExtract recurring user corrections into reusable guidance rules with frontmatter metadata.
GenEducationDetect deployment platforms and write deploy configuration into CLAUDE.md.
