Lightfast

The memory layer for software teams. Search everything your engineering org knows—code, PRs, docs, decisions—with answers that cite their sources.

Early Access — Lightfast is currently in early access. Request access →

Website · Documentation · Chat Demo · Discord

Why Lightfast?

Lightfast indexes code changes, PRs, docs, and decisions so engineers and AI agents can:

• Search by meaning — not just keywords
• Get answers with sources — trace decisions across your org
• Share memory — production-ready retrieval for agents

Supported Sources

Available now:

• GitHub — Pull requests, issues, code, discussions
• Vercel — Deployments, logs, project activity

Coming soon:

• Linear, Sentry, Slack, Notion, Confluence
• PlanetScale, Pulumi, Terraform, Zendesk
• Request an integration →

Use Cases

// "Why did we choose Postgres over MongoDB?"
await lightfast.search({ query: "database selection decision postgres mongodb" });

// "How does our payment integration work?"
await lightfast.search({ query: "stripe payment flow implementation" });

// "What broke in last week's deployment?"
await lightfast.search({ query: "production incident postmortem", filters: { dateRange: "7d" } });

// "Find PRs similar to this refactor"
await lightfast.findSimilar({ url: "https://github.com/org/repo/pull/123" });

// "What do we know about rate limiting?"
await lightfast.search({ query: "rate limiting implementation patterns" });

Example response:

{
  "results": [
    {
      "id": "doc_abc123",
      "type": "pull_request",
      "title": "Add rate limiting to API endpoints",
      "snippet": "Implemented token bucket algorithm with Redis...",
      "score": 0.92,
      "source": "github",
      "url": "https://github.com/org/repo/pull/456",
      "metadata": { "author": "jane", "mergedAt": "2024-12-01" }
    }
  ],
  "meta": { "total": 24, "latency": { "total": 145 } }
}

Security

• Your code stays yours — We index metadata and content for search, never train on your data
• Encrypted at rest and in transit — Industry-standard security practices
• Role-based access — Workspace permissions mirror your source permissions
• Self-hosted option — Coming soon for enterprises with strict data residency requirements

Requirements

• Node.js >= 18
• A Lightfast API key (request access)

Integrate in 2 Ways

1. TypeScript SDK

Install the lightfast package to add semantic search to any application:

npm install lightfast

import { Lightfast } from "lightfast";

// Pass API key directly or use LIGHTFAST_API_KEY environment variable
const lightfast = new Lightfast({ apiKey: process.env.LIGHTFAST_API_KEY });

// Search your workspace memory
const results = await lightfast.search({
  query: "how does authentication work",
  limit: 10,
});

// Get full document content
const content = await lightfast.contents({
  ids: ["doc_abc123"],
});

// Find similar documents
const similar = await lightfast.findSimilar({
  id: "doc_abc123",
  threshold: 0.7,
});

2. MCP Server (Claude, Cursor, Codex)

Connect AI assistants directly to your workspace memory via Model Context Protocol.

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "lightfast": {
      "command": "npx",
      "args": ["-y", "@lightfastai/mcp"],
      "env": {
        "LIGHTFAST_API_KEY": "sk-lf-..."
      }
    }
  }
}

Claude Code (CLI)

Add to .mcp.json in your project root (or ~/.claude.json for global):

{
  "mcpServers": {
    "lightfast": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@lightfastai/mcp"],
      "env": {
        "LIGHTFAST_API_KEY": "sk-lf-..."
      }
    }
  }
}

Or use the CLI: claude mcp add lightfast --scope project -- npx -y @lightfastai/mcp

Cursor

Add to .cursor/mcp.json in your project root:

{
  "mcpServers": {
    "lightfast": {
      "command": "npx",
      "args": ["-y", "@lightfastai/mcp"],
      "env": {
        "LIGHTFAST_API_KEY": "sk-lf-..."
      }
    }
  }
}

OpenAI Codex

Add to ~/.codex/config.toml:

[mcp_servers.lightfast]
command = "npx"
args = ["-y", "@lightfastai/mcp"]

[mcp_servers.lightfast.env]
LIGHTFAST_API_KEY = "sk-lf-..."

Available tools:

• lightfast_search — Search workspace memory
• lightfast_contents — Fetch full document content
• lightfast_find_similar — Find semantically similar documents

Get an API Key

1. Request early access to join the waitlist
2. Create a workspace and connect your sources (GitHub, docs, etc.)
3. Generate an API key from your workspace settings

Documentation

• API Reference — Full endpoint documentation
• SDK Guide — TypeScript SDK usage
• MCP Setup — Configure AI assistants
• Changelog — Latest updates and releases
• Blog — Tutorials, announcements, and deep dives

Contributing

We welcome contributions! See CONTRIBUTING.md for guidelines.

git clone https://github.com/lightfastai/lightfast.git
cd lightfast
pnpm install
pnpm dev

Community

• Discord — Chat with the team and community
• GitHub Issues — Report bugs and request features
• Twitter — Follow for updates

License

Lightfast uses a dual licensing approach:

All components are licensed under Apache License 2.0 — a permissive open source license.

For Users: You're covered by Apache-2.0 for all Lightfast components — use freely in commercial and non-commercial projects.

See LICENSING.md for complete details.