Developer Context Agent
A token-efficient, codebase-aware AI agent runtime that provides MCP tools for starting runs, getting run states, explaining code, and finding files.
README
Developer Context Agent
Token-efficient, codebase-aware AI agent runtime for TypeScript/Node.js developer workflows.
Overview
Developer Context Agent is a reusable backend service that acts as a context router and coding agent for IDEs, CLIs, and automation workflows. It minimizes token waste through:
- Hybrid retrieval (ripgrep + ts-morph symbols + import graph)
- Prompt-prefix caching
- Tool output summarization
- Token-budgeted context packages
- Supervised planner → executor → reviewer loop
Quick Start
Prerequisites
- Node.js 20+
- Docker (for Postgres + Redis)
- ripgrep (
rg) on PATH
Setup
cp .env.example .env
npm install
npm run docker:up
npm run db:migrate
npm run dev
Server runs at http://localhost:3100.
Groq (recommended for free tier)
- Get an API key from console.groq.com
- Add to
.env:
GROQ_API_KEY=gsk_your_key_here
DEFAULT_MODEL=groq:llama-3.3-70b-versatile
PLANNER_MODEL=groq:llama-3.1-8b-instant
Models use provider:model format — also supports openai: and anthropic: prefixes.
Health Check
curl http://localhost:3100/health
Explain a codebase (lightweight)
curl -X POST http://localhost:3100/chat \
-H "Content-Type: application/json" \
-H "X-API-Key: dev-key" \
-d '{"prompt":"What does auth do?","repoPath":"/path/to/your/repo"}'
Start a full run
curl -X POST http://localhost:3100/runs \
-H "Content-Type: application/json" \
-H "X-API-Key: dev-key" \
-d '{"mode":"find","prompt":"auth middleware bug","repoPath":"/path/to/your/repo"}'
API Endpoints
| Method | Path | Description |
|---|---|---|
| GET | /health |
Liveness + DB/Redis checks |
| GET | /metrics |
Token, cache, latency metrics |
| POST | /chat |
Lightweight explain mode |
| POST | /runs |
Start supervised agent run |
| GET | /runs/:id |
Get run state |
| POST | /runs/:id/approve |
Approve gated actions |
See docs/IDE_BRIDGE.md for IDE integration contract.
Architecture
src/
├── app/ # Fastify server + routes
├── core/ # Config, Zod schemas, types
├── orchestrator/ # Run lifecycle, planner, executor, reviewer
├── context/ # Repo map, retrieval, prompt building
├── tools/ # Sandboxed tool registry
├── memory/ # Multi-layer cache + session memory
├── models/ # OpenAI, Anthropic, local compressor
├── observability/ # Logging + metrics
└── mcp/ # MCP server for IDE integration
MCP Server
npx tsx src/mcp/server.ts
Tools: start_run, get_run, explain, find_files.
Testing
npm test
Unit tests run without Docker. Integration tests require Postgres + Redis.
Configuration
| Variable | Default | Description |
|---|---|---|
PORT |
3100 | HTTP port |
DATABASE_URL |
postgres://dca:dca@localhost:5432/developer_context_agent | Postgres |
REDIS_URL |
redis://localhost:6379 | Redis |
GROQ_API_KEY |
— | Groq API key (recommended for free tier) |
OPENAI_API_KEY |
— | OpenAI API key |
ANTHROPIC_API_KEY |
— | Anthropic API key |
DEFAULT_MODEL |
groq:llama-3.3-70b-versatile | Default model (provider:model) |
PLANNER_MODEL |
groq:llama-3.1-8b-instant | Planner model (provider:model) |
API_KEY |
— | Optional API auth |
TOKEN_BUDGET_DEFAULT |
32000 | Default token budget |
License
MIT
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
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.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.