Developer Edition — Free

Install Purple8

Purple8 Hyper Graph is your entire backend in one container — graph, vector, document & full-text storage, a built-in RAG pipeline, a 82-tool MCP server, workflows, auth, and encryption. No external services. The free Developer edition runs it all locally.

Recommended

One-click install

One command does everything: it installs Docker if you don’t have it, pulls the image, starts the container, and opens the console. Safe to re-run — it never deletes your data.

macOS & Linux

curl -fsSL https://www.purple8.ai/install.sh | bash

Windows (PowerShell)

irm https://www.purple8.ai/install.ps1 | iex

What the script does

Prefer to read before you run? View install.sh · View install.ps1. The script only pulls the public image — nothing phones home.

One container, your whole backend

Everything below runs inside the single purple8 container on port 8100 — replacing the stack you’d normally wire together yourself.

Prefer to do it by hand?

Three commands. You just need Docker 20.10+ (Docker Desktop on macOS/Windows, Docker Engine on Linux) and ~2 GB free RAM.

1. Pull the image

docker pull ghcr.io/purple8-technologies/purple8-graph:developer

2. Start the container

Port 8100 serves the REST API, MCP server, and admin console; the volume persists your data across restarts. Replace <your-license-key> with the PURPLE8_LICENSE_JWT from your activation screen or email.

docker run -d \
  --name purple8 \
  -p 8100:8100 \
  -e PURPLE8_LICENSE_JWT="<your-license-key>" \
  -v purple8-data:/data \
  ghcr.io/purple8-technologies/purple8-graph:developer

3. Open the console

Once the container is healthy, open the built-in admin console:

http://localhost:8100/lcnc

Check it’s running with docker logs -f purple8, or hit http://localhost:8100/health.

4. Connect an AI agent or coding tool

Purple8 exposes an MCP server at http://localhost:8100/mcp/sse. No separate SDK or plugin needed — any MCP-capable client connects with that URL and an API key. You can also call the 82 tools directly from your browser without any client at all.

Get your API key first — open the admin console at http://localhost:8100/lcnc/api-keys, create a key, and copy it. Paste it in place of YOUR_API_KEY in every snippet below.

Option A — Try it directly in your browser

The MCP Console is built into every Purple8 container. Open it, pick any of the 82 tools, fill in the arguments, and run — no client installation, no code.

http://localhost:8100/lcnc → MCP Console → Try It

Or call any tool directly over HTTP with curl:

curl -X POST http://localhost:8100/mcp \
  -H "X-API-Key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0",
    "method": "tools/call",
    "id": 1,
    "params": {
      "name": "graph.counts",
      "arguments": {}
    }
  }'

A successful response returns a JSON object with a result field containing node and edge counts. If you see that, the MCP server is live.

Option B — Claude Code (CLI)

One command registers Purple8 as an MCP server in your Claude Code session. After this, Claude can call all 82 tools from any conversation.

claude mcp add purple8 --transport sse http://localhost:8100/mcp/sse \
  --header "X-API-Key: YOUR_API_KEY"

Verify it worked: ask Claude “Call graph.counts and tell me how many nodes are in the database.”

Option C — Claude Desktop

Add Purple8 to your claude_desktop_config.json and restart Claude Desktop.

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "purple8-graph": {
      "url": "http://localhost:8100/mcp/sse"
    }
  }
}

Option D — Cursor

Add to your Cursor MCP config file (~/.cursor/mcp.json) or via Cursor Settings → MCP → Add Server:

{
  "mcp": {
    "servers": [
      {
        "name": "purple8-graph",
        "type": "sse",
        "url": "http://localhost:8100/mcp/sse",
        "headers": { "X-API-Key": "YOUR_API_KEY" }
      }
    ]
  }
}

Option E — VS Code / GitHub Copilot

Add to your VS Code .vscode/mcp.json (workspace) or settings.json (user-level):

{
  "mcp": {
    "servers": {
      "purple8-graph": {
        "type": "sse",
        "url": "http://localhost:8100/mcp/sse",
        "headers": { "X-API-Key": "YOUR_API_KEY" }
      }
    }
  }
}

After adding, open the Copilot Chat panel, switch to Agent mode, and Purple8 tools will appear in the tool list.

Option F — OpenAI Codex

Codex reads MCP server config from ~/.codex/config.json. Add the Purple8 server there:

{
  "mcpServers": {
    "purple8-graph": {
      "url": "http://localhost:8100/mcp/sse"
    }
  }
}

Restart Codex after saving. Purple8’s 82 tools will be available in every Codex session automatically. The same mcpServers key works for any OpenAI-compatible MCP host (ChatGPT Desktop, etc.).

Deploying to a server? Replace http://localhost:8100 with your host URL (e.g. https://my-purple8.fly.dev). All clients support remote URLs — no localhost tunnel required. Cloud deploy guide →

What’s in the Developer edition

• Graph + vector + document + full-text engine

• Built-in RAG pipeline (hybrid retrieval)

• MCP server — 82 tools across 11 namespaces

• Journey Engine (workflows, SLA, HITL, audit)

• LCNC admin console

• AES-256-GCM encryption at rest

• 50K nodes, 1 MCP agent (free tier)

• Perpetual license — no expiry

Optional

Add Purple8 DocIntel — turn any document into graph knowledge

DocIntel is a separate microservice that parses 70+ formats (PDF, Office, HTML, images, CAD & BIM), extracts entities and relationships, and emits them straight into your Purple8 graph. The Community edition is free — no license key, no expiry. It only needs an LLM API key for the extraction pipeline.

Run DocIntel (Community edition)

docker run -d \
  --name purple8-docintel \
  -p 8200:8200 \
  -e LLM__API_KEY="sk-..." \
  -e GRAPH__BASE_URL="http://host.docker.internal:8100" \
  ghcr.io/purple8-technologies/purple8-docintel:developer

Port 8200 serves the document API. GRAPH__BASE_URL points DocIntel’s emit callback at the Purple8 container you started above (host.docker.internal reaches your host from inside Docker Desktop; on Linux use --network host or the container IP). Check health at http://localhost:8200/health.

Send it a document

curl -X POST http://localhost:8200/process -F "file=@contract.pdf"

Extracted entities land in your graph automatically. Community edition covers 500 documents/mo on a single worker; paid tiers add throughput and managed connectors (SharePoint, Confluence, S3). See DocIntel plans.

Don’t have a license yet? Get one free in one click

Ready to scale beyond the free tier? See paid plans

Lost your key? Recover it here · Need help? support@purple8.ai