codebase-memory-mcp-intelligence

Codebase change impact analysis traces function calls by indexing source structure into a persistent knowledge graph, allowing you to query call chains and calculate risk scores for modifications. It maps dependencies to show exactly what breaks when code changes.

Dead code detection identifies unused functions by modeling your repository as a queryable graph of nodes and edges. By running graph queries against the indexed codebase, you can locate symbols with no incoming calls and safely remove them.

Semantic code search works without external API keys by using bundled embeddings to index and retrieve code concepts. It enriches the source structure graph so you can locate symbols and logic by meaning rather than exact string matching.

Yes, you can generate architecture summaries and list HTTP or gRPC routes for medium to large repositories. The codebase intelligence server indexes the source structure and extracts architectural boundaries and route definitions as queryable graph nodes.

No external API keys or dependencies are required to run codebase graph queries. The intelligence server uses bundled embeddings and local indexing, allowing you to execute Cypher-like queries and structural searches entirely offline.

A knowledge graph provides code intelligence by mapping functions, classes, and imports as structural nodes and edges, whereas standard text search only matches strings. This graph modeling enables deep call tracing, route discovery, and dependency analysis.