LocalSynapse Local File Search MCP Server
Let your coding agent search every local document offline with hybrid keyword and semantic ranking—no cloud upload.
Overview
LocalSynapse is an MCP server for the Build phase that provides offline hybrid BM25 and semantic search across local documents for AI coding agents.
What is this MCP server?
- Fully offline hybrid search combining BM25 keyword scoring with semantic embeddings
- Indexes and searches across all document types on the local machine
- Stdio MCP transport via packaged .mcpb release (v2.1.0)
- No API keys or network calls required for queries
- Pairs with agent workflows that need citations from local files only
- Server version 2.1.0 with stdio transport
- Search stack: BM25 plus semantic hybrid (offline)
- Distributed as registryType mcpb with published fileSha256
What problem does it solve?
Agents only see what you paste or what is in git, so critical context locked in local PDFs and notes stays invisible during coding sessions.
Who is it for?
Privacy-conscious solo builders with large offline archives who want semantic file search inside the agent loop.
Skip if: Teams that need collaborative cloud search, live web research, or querying code-only repositories without document exports.
What do I get? / Deliverables
Your agent can query indexed local files and return ranked snippets you can cite while staying completely offline.
- Hybrid-ranked search results from local indexed documents
- Agent-callable offline search tools over your file corpus
- Citable excerpts pulled from files outside git context
Recommended MCP Servers
Journey fit
While you build, specs, notes, and past exports pile up on disk; this server is shelved under build/docs because agents need fast retrieval of project knowledge without leaving the IDE. Docs subphase covers internal wikis, READMEs, exports, and scratch notes that are not in git context windows—exactly what LocalSynapse indexes and ranks.
How it compares
Local file retrieval MCP server, not a hosted vector database or browser automation skill.
Common Questions / FAQ
Who is LocalSynapse for?
It is for indie builders and developers who keep specs, research, and exports on disk and want their AI agent to search those files without uploading them anywhere.
When should I use LocalSynapse?
Use it during build and docs work whenever you need answers from local PDFs, notes, or exports that are not in the current repo or chat context.
How do I add LocalSynapse to my agent?
Register the stdio MCP package from the v2.1.0 .mcpb release in Claude Desktop, Claude Code, or Cursor MCP settings, then configure which local paths to index.