Search meaning, not code chunks

THINKER.md transforms your repository into a versioned, LLM-friendly contextual layer. Enable AI agents to understand intent, boundaries, and relationships — reducing noise, cost, and hallucinations.

Get StartedLearn More

From code to context

Your repository transforms into an LLM-friendly meta-repository.Context files describe purpose, responsibilities, and relationships.

Connecting to repository
Analyzing file structure
Extracting intent and boundaries
Generating context files
Syncing to cloud
Meta-repository ready
meta-repository
repo.context.md
src/
src.context.md
assets/
assets.context.md
logo.png.context.md
page.tsx.context.md
route.ts.context.md
public/
public.context.md
file.pdf.context.md

The difference is clear

Same query, dramatically different results

semantic_search
inefficient
Query:"how does authentication work"
Searching codebase...
Results: 47 chunks retrieved
src/auth/middleware.ts:1-89
import { NextRequest } from 'next/server' import { verify } from 'jsonwebtoken' export async function middleware(req...
src/lib/auth.ts:23-156
export const authOptions: AuthOptions = { providers: [ CredentialsProvider({...
src/components/LoginForm.tsx:1-203
'use client' import { useState } from 'react'...
... 44 more chunks
~34,000 tokens consumed
Relevance: ~40%
mcp_thinkermd_search
optimized
Query:"how does authentication work"
Fetching context...
Results: 3 context files retrieved
auth.context.md
Purpose: Handles user authentication and session management using JWT tokens with HTTP-only cookies.
Flow: Login → Validate credentials → Issue JWT → Store in cookie → Middleware validates on each request.
Dependencies: next-auth, jsonwebtoken, bcrypt
Related: middleware.context.md, user.context.md
middleware.context.md
Purpose: Protects routes by validating JWT tokens before allowing access.
Boundary: Only handles token validation, does not manage sessions or user data.
~1,200 tokens consumed
Relevance: ~95%

28x fewer tokens. 2x more relevant. Reduced hallucinations.

Always in sync

Context updates automatically with every push to GitHub.

a3f2c1d
8b4e2a1
c7d3f9e
1e5a8b2
code changes
Select a commit...
context update
Waiting...

Every commit. Every branch. Context that evolves with your code.

Built for developers

Powerful API to build the next generation of AI coding agents

terminal
# Search context via REST API
curl -X POST https://api.thinkermd.com/v1/search \
  -H "Authorization: Bearer your-key" \
  -H "Content-Type: application/json" \
  -d '{
  "query": "authentication flow",
  "repo": "your-org/your-repo",
  "branch": "main"
}'

REST API

Simple HTTP endpoints for any language or platform

Official SDKs

TypeScript, Python, and Go libraries ready to use

Build agents

Create AI agents with deep codebase understanding

Enterprise ready. Custom integrations, dedicated support, and SLAs for teams building at scale.

Pricing

Start free, scale when you need

Free

$0forever

For individuals and small projects

  • 1 meta-repository
  • Manual sync
  • Community support

Pro

$15/month

For professionals and teams

  • 3 meta-repositories
  • Auto-sync on push
  • Branch syncing
  • MCP integration

Enterprise

Custom

For teams building with API

  • Everything in Pro
  • REST API access
  • Official SDKs
  • Custom integrations
  • Dedicated support
  • SLA guarantee

FAQ

Common questions about THINKER.md