oops - Handle mistake corrections and generate preventive guard scripts
Parses user-reported corrections, extracts structured learnings, records feedback to auto-memory, and generates guard scripts to prevent recurring mistakes.
Tags
Updated: 2026-06-30Capabilities
Typical Inputs
Typical Outputs
What this skill does
- parse correction size from message
- extract structured learning from mistake
- run correction shell script
- write learning to auto-memory file
- append learning to global feedback file
- determine learning scope
- generate guard script
- create trigger metadata file
- self-test guard script
- report correction results
Inputs
- User correction message
- Recent conversation context
- Correction shell script path
- Auto-memory directory path
- Global feedback file path
- Guard script directory path
Outputs
- Structured learning entries in respect-feedback.md
- Global feedback entries in global-feedback.md
- Guard script files
- Trigger JSON files
- Correction status messages
- Error messages on script failure
Requirements
- Bash execution environment
- correct.sh script at ~/.claude/respect/plugin_root/scripts/
- Read/write access to ~/.claude/ directory tree
- Auto-memory directory structure under ~/.claude/projects/