symapse
Provides code analysis tools for AI coding agents to understand project architecture, search for symbols, and plan changes with token reduction.
README
Symapse
Architectural awareness engine for AI coding agents
<p align="center"> <img alt="npm version" src="https://img.shields.io/npm/v/symapse?color=blue"> <img alt="license" src="https://img.shields.io/npm/l/symapse"> <img alt="node" src="https://img.shields.io/node/v/symapse"> </p>
Benchmarks
Same prompt, same codebase. One agent uses Symapse, one reads files directly.
| Repo | Language | Symapse | Control | Reduction |
|---|---|---|---|---|
| Django | Python | 57k | 310k | −81% |
| React | JavaScript | 40k | 80k | −50% |
| Prettier | JavaScript | 49k | 90k | −46% |
| Fzf | Go | 38k | 68k | −44% |
| FastAPI | Python | 51k | 70k | −27% |
| SQLite | C | 44k | 44k | 0% |
Average: 50% token reduction. Plan quality: consistently 4/5 vs 5/5.
<p align="center"><img src="docs/benchmarks.png" alt="Symapse Benchmarks" width="720"></p>
Install
npm install -g symapse
Requires Node.js 22+. No dependencies. Zero config.
Quick Start
# Index your project
symapse index /path/to/repo
# Start the MCP server (for OpenCode/Cursor/Claude integration)
symapse mcp /path/to/repo
# Query from the terminal
symapse architecture
symapse impact "login"
symapse deadcode 10
symapse where "add notification system"
symapse context "direct login after payment"
OpenCode Integration
Install Symapse first:
npm install -g symapse
Add to your project's opencode.json:
{
"mcp": {
"symapse": {
"type": "local",
"command": ["npx", "symapse", "mcp", "."],
"enabled": true
}
}
}
Create an AGENTS.md in your project root with:
## RULE 1: First action — symapse_clarify
Before ANYTHING else, call `symapse_clarify "<request>"`. If it returns questions, ASK them. Do not guess.
## RULE 2: Always check Symapse before reading files
| Instead of... | Use... |
|---|---|
| Reading files to understand repo | `symapse_architecture` |
| Grepping for symbols | `symapse_search` |
| Tracing callers | `symapse_impact` |
| Guessing where code goes | `symapse_where` |
| Checking for duplicates | `symapse_overlap` |
| Finding dead code | `symapse_deadcode` |
| Finding must-read files | `symapse_context` |
If Symapse doesn't answer your question, read files. But check Symapse first.
Tools
| Tool | Question it answers |
|---|---|
symapse_architecture |
Explain this repo |
symapse_clarify |
What am I assuming? |
symapse_search |
Where does this thing live? |
symapse_impact |
What breaks if I change this? |
symapse_deadcode |
What can I safely delete? |
symapse_overlap |
Did we build this already? |
symapse_where |
Where should this new thing go? |
symapse_context |
Which files must I read? |
symapse_conventions |
What patterns should I follow? |
symapse_changes |
What changed? |
Supported Languages
JavaScript, TypeScript, Python, Go, Rust, C#, PHP, Ruby, Lua, and C.
How It Works
- Index — walks the repo, extracts functions/classes/methods with regex parsers, builds call and import edges
- Store — normalized SQLite schema, incremental by mtime + engine version
- Expose — CLI, REST API, and MCP over stdio. All share the same engine.
Self-Improving
Symapse learns from agent behavior across sessions without any user involvement:
- Session coherence — biases context toward the subsystem the agent is already exploring
- Usage signals — logs which symbols agents actually drill into
- Workflow memory — auto-detects repeated symbol sequences across sessions
- Intent classification — routes config/docs questions away from source code exploration
Architecture
packages/
engine/ — regex-based parser, call graph, all analysis tools
db/ — SQLite persistence with incremental indexing
mcp/ — MCP protocol server (JSON-RPC over stdio)
apps/
cli/ — command-line entry point
api/ — HTTP API server + web dashboard
web/ — terminal-style web UI
License
AGPL-3.0
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.