scribe

Documents software repositories into Claude-facing files from verified facts.

2|Updated Feb 23, 2026
One-click install
npx skills add https://github.com/MDGrey33/memnyx --skill scribe-mdgrey33
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: scribe
Source: https://github.com/MDGrey33/memnyx/tree/main/.claude/skills/scribe
Command: npx skills add https://github.com/MDGrey33/memnyx --skill scribe-mdgrey33

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Scribe turns a software repository into trustworthy, Claude-facing documentation so future work does not depend on tribal knowledge or stale prose. It helps prevent confidently wrong docs by forcing every important claim to be grounded in source.

Core Features & Use Cases

  • Verified documentation generation: Creates or updates CLAUDE.md, reference docs, project context, and README content from facts that can be checked in the repository.
  • Doc-state detection: Detects whether a repo needs fresh documentation, an update, or a partial fill-in of missing structure.
  • Safety for uncertain claims: Separates confirmed facts from unverified or contradictory details so hazards do not get silently written into committed docs.
  • Use case: A team inherits an unfamiliar codebase and uses Scribe to produce a clean documentation layer that explains setup, structure, and operating conventions without guessing.

Quick Start

Ask Scribe to document the project from verified facts and create the missing Claude-facing documentation files.

Frequently Asked Questions about scribe

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate verified documentation from an existing codebase?

Generate verified documentation by scanning the repository source code to extract confirmed facts, which prevents confidently wrong or stale project knowledge from being committed. The process creates CLAUDE.md files and reference docs grounded in actual source code.

What is the best way to keep CLAUDE.md project context files from getting stale?

Keeping CLAUDE.md files from getting stale requires detecting the current doc-state of the repository and syncing updates only from verified source facts. This separates confirmed details from unverified claims to ensure documentation remains accurate.

How does repository analysis separate confirmed facts from uncertain claims in knowledge bases?

Repository analysis separates confirmed facts from uncertain claims by enforcing source-grounded verification and tracking hazards for contradictory details. This prevents unverified information from being silently written into committed markdown knowledge bases.

Can I use markdown documentation workflows to fill in missing repository READMEs?

You can use markdown documentation workflows to fill in missing repository READMEs by detecting partial or missing structure in the project. The workflow preserves existing doc-layers while generating new content strictly from verified repository facts.

When do I need source-grounded verification for creating reference docs?

Source-grounded verification for creating reference docs is needed when inheriting an unfamiliar codebase or updating operating conventions. It ensures every important claim is checked against the repository to avoid tribal knowledge dependencies.

What are the limitations of generating knowledge management files without hazards tracking?

Generating knowledge management files without hazards tracking risks silently writing unverified or contradictory details into committed documentation. Without separating confirmed facts, future work depends on misleading prose rather than grounded repository truths.