Log mistakes, corrections, and patterns to .scratch/napkin.md.
npx skills add https://github.com/sandover/codex-skills --skill napkin
Please help me install this Agent Skill. Skill: napkin Source: https://github.com/sandover/codex-skills/tree/main/skills/napkin Command: npx skills add https://github.com/sandover/codex-skills --skill napkin
Maintain a per-repo napkin file that tracks mistakes, corrections, and what works. This living document activates every session, unconditionally. Read the napkin before doing anything, and write to it continuously as you work — not just at session boundaries. Log your own mistakes, not just user corrections. The napkin lives in the repo at .scratch/napkin.md.
Create or open the file .scratch/napkin.md in your repository at the start of each session and begin logging notes continuously.
High-intent search queries and answers about installing and using this skill.
A living document tracks mistakes and corrections by logging them continuously into a structured in-repo file at .scratch/napkin.md, preserving context across sessions for continuous improvement.
The best way to maintain continuous improvement notes is using a living document activated at session start that organizes insights into structured sections for Corrections, User Preferences, and Patterns within the repository.
Log session insights by opening a running notes file at session start and updating it continuously during work, recording mistakes, corrections, and patterns in structured markdown sections for ongoing reference.
Yes, you can customize the documentation sections. The structured sections for Corrections, User Preferences, and Patterns can be redesigned to fit your specific repository domain and workflow requirements.
No, the napkin approach requires zero external dependencies. Notes are stored locally as a markdown file within your repository, leveraging standard version control for tracking and collaboration.
Avoid using a living document for learning if your workflow cannot accommodate continuous updates during sessions. The system relies on unconditionally reading the file at start and writing to it continuously during work.
KanevryInitialize and analyze git repository state and project readiness for engineering sessions.
parcadeiConvert session learnings into reusable skills, rules, and agents.
theaichimeraSave conversation insights as YAML-frontmatter Markdown skill files in a Git-backed project repository.
LuminaVaultCompile raw markdown files into structured summaries and concept articles.
