Lumiqa MCP Docs

Get started in 3 steps

What is MCP?

MCP (Model Context Protocol) is Anthropic's open standard for connecting AI agents to external tools and data. Instead of writing custom integrations for every platform, an MCP-capable client (Claude Desktop, Claude Code, ChatGPT with the right plugin, your own agent) discovers tools and calls them through a single JSON-RPC interface. Lumiqa was built MCP-first: every action a human can do in the dashboard is also a typed, documented tool that an agent can invoke. Read the full spec at modelcontextprotocol.io.

The 21 tools, by category

Connect with Claude Desktop

One command and Claude can manage the entire workspace. Replace <your-slug> with your workspace slug (visible at the bottom of the dashboard) and paste a fresh API key when prompted.

shell# Add Lumiqa as an MCP server in Claude Desktop / Claude Code
$ claude mcp add lumiqa https://lumiqa.io/mcp/<your-slug> \
    --header "Authorization: Bearer lk_live_xxxxxxxxxxxx"

# Verify it shows up
$ claude mcp list
lumiqa  21 tools  ·  connected

Now you can just ask Claude things like:

prompt"List everything waiting for approval, run a compliance scan on each one,
 then approve the ones with no high-risk violations and deliver them."

Call via REST

If you'd rather skip MCP, every tool is also reachable over plain HTTPS. The endpoint is POST /mcp/<workspace-slug> and the body is a JSON-RPC 2.0 envelope.

Discovery (GET) — no auth

shell$ curl https://lumiqa.io/mcp/<your-slug>
# Returns the full TOOLS manifest as JSON

List tools (POST tools/list)

shell$ curl -X POST https://lumiqa.io/mcp/<your-slug> \
    -H "Authorization: Bearer lk_live_xxxxxxxxxxxx" \
    -H "Content-Type: application/json" \
    -d '{
      "jsonrpc": "2.0",
      "id": 1,
      "method": "tools/list"
    }'

Call a tool (POST tools/call)

shell$ curl -X POST https://lumiqa.io/mcp/<your-slug> \
    -H "Authorization: Bearer lk_live_xxxxxxxxxxxx" \
    -H "Content-Type: application/json" \
    -d '{
      "jsonrpc": "2.0",
      "id": 2,
      "method": "tools/call",
      "params": {
        "name": "list_content",
        "arguments": { "status": "Attesa Approvazione", "limit": 10 }
      }
    }'

Responses follow the JSON-RPC shape:

json{
  "jsonrpc": "2.0",
  "id": 2,
  "result": {
    "content": [
      { "type": "text", "text": "{ \"items\": [...] }" }
    ]
  }
}

Authentication & API keys

Every write request requires a workspace-scoped API key. Keys start with lk_live_ and are tied to a single workspace slug — using a key against the wrong slug returns 403 workspace_mismatch.

• Get a key: Settings → API Keys → Generate. Keys are shown once; copy immediately.
• Scope: default scope is read+write. Read-only keys cannot call any write tool (see per-tool docs for the write badge).
• Rotation: revoke from the same dashboard; existing keys keep working until you click Revoke.
• Header: Authorization: Bearer lk_live_xxxxxxxxxxxx

Versioning

v1.0 — pivot 2026-05-21 The 21-tool surface frozen in this doc set is the current generally-available version. Breaking changes will ship as v2 on a parallel route (/mcp/v2/<slug>). Additive changes — new tools, new optional arguments, new response fields — can land at any time and are documented in the changelog.

Error shape

All tool errors land as standard JSON-RPC errors:

json{
  "jsonrpc": "2.0",
  "id": 2,
  "error": { "code": -32000, "message": "Tool execution failed" }
}

Common codes:

• -32700 — malformed JSON
• -32600 — missing or wrong jsonrpc version
• -32601 — unknown method
• -32000 — tool execution failed or insufficient scopes
• 401 (HTTP) — missing / invalid / expired API key
• 403 (HTTP) — key's workspace slug doesn't match the route

Next

Pick a tool from the sidebar to see its schema, examples, and limitations. Most users start with upload_video → request_review → approve_content → deliver_asset, and graduate to run_workflow once they want one-call macros.