code-comprehension

A Claude skill that tours a codebase you built with AI until you can explain how it works: traces one real run, names your patterns, makes you say it back. Live tour, print workbook, or grade mode. Companion to explain-diff.

prints csv tree as a tree, much like cflow

Trace before change. Reconcile after change. A semantic layer that lets coding agents see a system as behavior — trace what a change might affect before making it, and check the model is still true after.

读懂代码的坐标系 · The coordinate system for code explanations

Incomplete? I think this is challenging to implement

Agent Skill: whole-system codebase comprehension for your AI agent, powered by peek (codepeekr.dev)

Run several Claude Code agents on one repo. 'Done' means the project's real checks actually passed, not that an agent said so.

Version 1.0 from 2012

Understand an unfamiliar codebase: reading order, ownership, change coupling, hotspots and findings computed from the code and its git history, with Q&A whose citations are checked.

Human-controlled AI development: understand the landscape, choose the path, and steer the work.

Tracks the code Claude wrote that you never understood. Charges interest. Makes you explain it back before it compounds.

A Haskell plugin that is used to parse Haskell projects for codecompass. The data necessary for providing code comprehension for the project is exported.

Official MINIX sources - Automatically replicated from gerrit.minix3.org

Swamp extension that turns a Git repo or an architecture diagram into an illustrated, wikilinked Obsidian atlas explaining the system.

version 2.0 (2016)

A swarm of specialist clones that understand a codebase, then guide every move the LLM operator makes. MiroThinker architecture ported to code. OpenHands sub-agent/plugin.

Combat cognitive debt from Claude AI-assisted development

A Claude skill for reading a codebase you did not write. It keeps the agent's answers short, plain and complete, so you can join a project in the middle and understand it.

Benchmark for repository-level code understanding: 500 open-ended questions from real GitHub Discussions across 50 repositories, graded by a checklist-based verification protocol (ASE 2026)

Understand unfamiliar code with step-by-step walkthroughs and follow-up questions.