Single-binary graph storage
The graph lives locally, uses an embedded SQLite-compatible backend, and does not need a hosted graph service to start working.
raph indexes local code, docs, and scoped memory into a graph under ~/.raph/, then exposes that graph through MCP and Studio so coding agents can search, traverse, and keep context close to the repo.
The graph lives locally, uses an embedded SQLite-compatible backend, and does not need a hosted graph service to start working.
Agents can store scoped memory, run hybrid search, index codebases, crawl docs, and walk related nodes from the same local store.
Open the local UI to inspect nodes, tables, metadata, neighbor relationships, and maintenance actions without leaving your machine.
The docs are now broken into topic pages so each area can be more descriptive without forcing readers through one long scroll. Use the pages below as your entry points.
Install raph, initialize config, index a repository, start the MCP server, and open Studio.
See the MCP tools, supported agent config targets, and the exact config shape each agent expects.
Understand the graph explorer, local maintenance actions, and what to expect when inspecting nodes and tables.
Use the command cheat sheet, install options, update behavior, and quick fixes for common setup issues.
cmd/raph.internal/config.internal/db.internal/indexer.internal/syncer.internal/mcp.internal/studio.The first implementation stays local-first with a pure-Go embedded backend so macOS, Linux, and Windows builds work cleanly from day one.