Living Codebase Cartographer — AI skill for Claude Code

Living Codebase Cartographer — offline codebase map, dependency graph & impact analysis (no LSP, no cloud)

**Map any codebase offline in seconds: endpoints, call graphs, data lineage, blast-radius (impact) analysis, and request-flow tracing — with evidence + confidence on every fact. Zero dependencies, zero network calls, zero telemetry.** Python stdlib only, works without an LSP or language server, generates a portable offline HTML architecture explorer that opens over `file://`.

A **language-agnostic, framework-agnostic living codebase intelligence system**. It builds and maintains a machine-readable map of any software repository — symbols, API endpoints, call graphs, data stores, dependencies, call edges — with evidence + confidence on every fact, and keeps it synchronized as the code evolves. Think: an **offline alternative** to cloud code-graph tools, an **AI-agent skill for repo mapping**, and a **standalone static-analysis CLI** in one package.

**Keywords:** codebase map generator, repository mapping tool, static code analysis python, offline code intelligence, dependency graph visualizer, call-graph generator, blast-radius / impact analysis, API endpoint inventory, data-lineage tracker, architecture explorer, monorepo documentation generator, AI coding-assistant skill, DeepSeek / Claude / agentic-coding skill, no-LSP code navigation, `file://` HTML code browser.

Supported out of the box: Java, TypeScript/JavaScript, Python, Go, C#, Rust, SQL (any DDL dialect), manifests/config/CI (generic), plus a LOW-confidence fallback pass for anything else (Ruby, PHP, Kotlin, Swift, …). Technology identity lives in `meta` (`lang`, `framework`, `dialect`); the graph schema itself is closed and technology-independent.

Requirements

• Python 3.8+ (stdlib only — no third-party dependencies)
• git (for change detection; works without git, using file hashes)

Install

Clone the repo, then copy it into your project as a skill:

git clone http