What is the primary benefit of the Codification Loop?

It solves the problem of AI agents losing context between sessions by capturing breakthroughs and decisions into durable markdown files that provide context for all future sessions.

How does this help with token efficiency?

By documenting recurring information, the AI can reference concise local files instead of requiring long, repetitive explanations in the chat history, saving tokens and reducing noise.

What kind of information should I codify?

You should codify anything you've explained twice, complex debugging solutions, architectural trade-offs, non-obvious 'gotchas,' and newly established project conventions.

Does this skill modify my code automatically?

No. The Codification Loop follows a strict 'Approve' phase where it shows you the proposed documentation update and requires your confirmation before writing to any file.

Where does the captured knowledge get stored?

Knowledge is routed to specific local documentation files such as constitution.md for rules, failure-modes.md for bugs, and specs/ directories for architectural decisions.

Codification Loop

Name: Codification Loop
Author: pwarnock

bypwarnock

0•

Productivity & Workflow

Captures ephemeral debugging insights and architecture decisions into durable project context to eliminate AI memory loss between sessions.

The Codification Loop is a specialized Claude Code skill that implements the 'explained it twice, write it down' methodology to solve the problem of AI 'amnesia' across coding sessions. It empowers the AI agent to self-monitor for high-value knowledge—such as debugging breakthroughs, design trade-offs, and new conventions—and propose drafting them into persistent markdown files like a project constitution or failure modes log. By turning fleeting conversation details into structured, durable documentation, this skill ensures that project context accumulates over time, reducing redundant explanations and preventing the re-discovery of previously solved bugs.

Key Features

01Context-aware drafting that matches existing project documentation styles

02Standardized entry templates for conventions, gotchas, and architecture decisions

03Automatic signal detection for repeated explanations and debugging breakthroughs

040 GitHub stars

05Human-in-the-loop approval workflow to maintain documentation quality

06Intelligent knowledge routing to specific files like constitution.md and failure-modes.md

Use Cases

01Maintaining an up-to-date failure-modes log for complex environment-specific issues

02Codifying team-specific naming conventions and architectural standards in real-time

03Preserving non-obvious debugging insights to prevent future regressions

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add pwarnock/pwarnock-cc-plugins codification-loop

For use in Claude.ai and ChatGPT

Key Features

01Context-aware drafting that matches existing project documentation styles

02Standardized entry templates for conventions, gotchas, and architecture decisions

03Automatic signal detection for repeated explanations and debugging breakthroughs

040 GitHub stars

05Human-in-the-loop approval workflow to maintain documentation quality

06Intelligent knowledge routing to specific files like constitution.md and failure-modes.md

Use Cases

01Maintaining an up-to-date failure-modes log for complex environment-specific issues

02Codifying team-specific naming conventions and architectural standards in real-time

03Preserving non-obvious debugging insights to prevent future regressions