BACK_TO_FEEDAICRIER_2
GitNexus builds knowledge graph for AI coding agents
OPEN_SOURCE ↗
GH · GITHUB// 28d agoOPENSOURCE RELEASE

GitNexus builds knowledge graph for AI coding agents

GitNexus is a zero-server code intelligence engine that indexes any codebase into a persistent knowledge graph — tracking every dependency, call chain, and functional cluster — and exposes it to AI coding agents via MCP tools. It runs entirely in-browser via WebAssembly or as a local CLI, giving agents like Cursor, Claude Code, and Windsurf structural awareness before they make changes.

// ANALYSIS

This is the missing context layer that AI coding agents have desperately needed — blast-radius analysis and dependency graphs that prevent agents from shipping blind edits.

  • Precomputes structure at index time (clusters, blast radii, execution flows) rather than making agents iteratively explore raw graph edges — one tool call returns decision-ready context
  • 7 MCP tools cover impact analysis, hybrid search (BM25 + semantic + RRF), coordinated renaming, and raw Cypher queries against the knowledge graph
  • Browser mode via Tree-sitter WASM + KuzuDB WASM is genuinely clever — full codebase indexing with no install, no server, no API keys for repos up to ~5k files
  • Supports 13 languages and 30+ node types; the 9-phase ingestion pipeline is unusually thorough for an OSS project from a solo developer
  • 13k+ stars acquired since August 2025 signals strong developer appetite for agent infrastructure tooling beyond the LLM itself
// TAGS
gitnexusagentmcpragdevtoolopen-sourcegraph

DISCOVERED

28d ago

2026-03-15

PUBLISHED

28d ago

2026-03-15

RELEVANCE

8/ 10