ratchet-map

Generates a coarse, language-agnostic architecture map of a codebase into memory/ARCHITECTURE.md.

2|Updated Jun 13, 2026
One-click install
npx skills add https://github.com/praveenvijayan/Ratchet --skill ratchet-map-praveenvijayan
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ratchet-map
Source: https://github.com/praveenvijayan/Ratchet/tree/main/plugin/skills/ratchet-map
Command: npx skills add https://github.com/praveenvijayan/Ratchet --skill ratchet-map-praveenvijayan

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When a codebase drifts over time, an agent's mental model of the project structure becomes stale, leading to wasted reads and wrong assumptions. This Skill regenerates a fresh, coarse architecture map so the agent can orient itself and scope its reads accurately. ## Core Features & Use Cases - Language-agnostic project detection: Identifies the ecosystem from manifests like package.json, pyproject.toml, Cargo.toml, go.mod, and pom.xml, then locates conventional source roots. - Coarse structural mapping: Describes top-level directories, major components by role, and project conventions while ignoring generated and vendor directories such as node_modules, dist, and target. - Review-safe output: Writes only memory/ARCHITECTURE.md, marks it machine-generated and provisional, and stops without committing so a human can review the diff. - Use Case: After a large refactor moves modules around, run this Skill to rebuild the architecture map so subsequent agent sessions navigate the new layout correctly. ## Quick Start Regenerate the architecture map of this repository into memory/ARCHITECTURE.md and stop for my review.

Frequently Asked Questions about ratchet-map

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

FAQPage Schema
How do I generate an architecture map of my codebase for an AI agent?

Run the ratchet-map skill, which detects your project type from manifest files, walks the source directories, and writes a coarse map to memory/ARCHITECTURE.md. It stops after writing so you can review the diff before committing.

What languages and project types does codebase mapping support?

The mapping is language-agnostic. It detects ecosystems from manifests such as package.json, pyproject.toml, Cargo.toml, go.mod, pubspec.yaml, pom.xml, and Gemfile, and falls back to describing top-level directories for unrecognized stacks.

Does the architecture map include generated or vendor directories?

No. The skill explicitly ignores generated and vendor directories such as build, dist, target, node_modules, vendor, __pycache__, and .next, and never reads into them.

Will the skill commit the generated ARCHITECTURE.md file?

No. It writes only memory/ARCHITECTURE.md and then stops, leaving the diff for human review. It never commits, pushes, or touches any other file, code, or branch.

How detailed is the generated architecture map?

The map is deliberately coarse: directories, responsibilities, and components by role only. It excludes line numbers, function signatures, dependency versions, and absolute paths, and the code remains the authority when they disagree.