codebase-memory-mcp-intelligence

Indexes codebases into knowledge graphs for structural queries, call tracing, and architecture analysis.

7|2|Updated May 16, 2026
One-click install
npx skills add https://github.com/reason-machines/mcp-skills --skill codebase-memory-mcp-intelligence-reason-machines
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: codebase-memory-mcp-intelligence
Source: https://github.com/reason-machines/mcp-skills/tree/main/skills/codebase-memory-mcp-intelligence
Command: npx skills add https://github.com/reason-machines/mcp-skills --skill codebase-memory-mcp-intelligence-reason-machines

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Understanding large or unfamiliar codebases requires hours of manual reading, and answering questions like "what calls this function" or "what breaks if I change this" is slow and error-prone. This Skill indexes codebases into a persistent knowledge graph so agents can answer structural questions in milliseconds. ## Core Features & Use Cases - Graph-Based Code Intelligence: Parses 155 languages with tree-sitter into queryable nodes and edges covering functions, classes, calls, imports, and inheritance. - Call Tracing & Impact Analysis: Finds all callers of a symbol, traces execution chains, detects dead code, and classifies the risk of uncommitted git changes. - Architecture & Semantic Search: Generates architecture summaries with hotspots and module boundaries, plus vector search using bundled embeddings with no API key. - Use Case: When onboarding to a new repository, ask the agent to index the project, show the architecture, list HTTP routes, and identify hotspot functions to build a mental model in minutes. ## Quick Start Ask your agent to index this project and show me the architecture of the codebase.

Frequently Asked Questions about codebase-memory-mcp-intelligence

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

FAQPage Schema
How do I index a codebase for structural search?▼

Ask your agent to index the project, which calls the index_repository tool on the current working directory, or run codebase-memory-mcp index /path/to/repo from the CLI. You can also enable auto_index so new projects index automatically on MCP session start.

How to find all callers of a function in a large codebase?▼

Use the get_callers tool with the symbol name and a max_depth parameter to traverse incoming CALLS edges in the knowledge graph. For deeper analysis, trace_symbol follows call chains in either direction up to a configurable depth.

What languages does tree-sitter code indexing support?▼

The server parses 155 languages using tree-sitter grammars compiled into a single static binary. Language-specific features include Go package resolution, TypeScript JSX tracing, Python decorator route detection, and C/C++ header linking.

Does semantic code search require an API key?▼

No, semantic search uses bundled Nomic embeddings that run locally without any API key. It combines 11 scoring signals including TF-IDF relevance, AST structural similarity, and MinHash near-clone detection.

Why is indexing slow or timing out on large repositories?▼

Large repos may exceed the default 8GB memory limit or include oversized files. Increase max_memory_gb, lower max_file_size_mb, and add a .cbmignore file to exclude directories like node_modules, dist, and vendor.

Can I detect the impact of uncommitted code changes?▼

Yes, the detect_changes tool analyzes your git diff and returns changed symbols with confidence scores, affected downstream symbols via the call graph, and a risk classification of CRITICAL, HIGH, MEDIUM, or LOW.