ADOS

A Model Context Protocol server for the ADOS drone platform. Read and control a drone or a whole fleet from your AI client, with a scope on every connection, a confirmation on anything that moves an aircraft, and an audit trail on every call.

License: MIT TypeScript MCP PRs Welcome

Part of the ADOS ecosystem. Pairs with the ADOS Drone Agent, ADOS Mission Control, and ADOS Extensions.

Docs · Issues · Model Context Protocol

What it is

ADOS MCP lets any MCP-capable AI client, Claude Code first-class, plus Claude Desktop, Cursor, VS Code, or a plain HTTP client, connect to an ADOS drone or ground station and:

• Read live status, telemetry, and health for a node or a fleet.
• Read every flight-controller parameter with full meaning and get tuning recommendations.
• Administer the platform: rename a node, update the software, install and configure plugins, restart a service, manage pairing and network settings.
• Query logs and a full audit trail.
• Fly, when the operator explicitly grants it: arm, change mode, go to a point, run a mission, land, return home, or stop.

The drone is the server; the AI is the client. No model runs on the drone. The server is a thin layer over interfaces that already exist, so behavior is defined once and stays consistent whether it is reached from the console, the command line, or an AI client.

Visibility is broad; action is gated. Reading is open within the granted scope. Writing is controlled by that scope, a safety class per action, a confirmation step for anything that moves an aircraft or changes state, and an operator-present check for flight. Every call, allowed or denied, is recorded.

Two ways to run it

You run the server yourself, on your own machine, in one of two modes:

Both expose the identical tools; only the reach differs.

• Fleet-mode is the primary pathway. The server connects to a Mission Control (GCS) backend — your own local one (--gcs local) or the production one (--gcs prod) — as you, the operator, and reaches the drones that Mission Control manages. It is the AI-native interface to your fleet. (Reach limit: Mission Control tracks cloud-connected drones; a drone that is only on your LAN is reached in agent-mode.)
• Agent-mode is the local-first direct pathway. The server points straight at one drone's agent on the LAN — no cloud round-trip — the shape a bench or field setup uses.

Quick start (Claude Code)

Sign in and mint one scoped machine credential in the Mission Control MCP tab, then run the server on your machine:

# Your whole fleet, through Mission Control (production backend):
export ADOS_MCP_TOKEN="paste-the-machine-credential-from-the-tab"
claude mcp add ados -- npx -y @altnautica/ados-mcp --target fleet --gcs prod

# One drone, directly on the LAN:
claude mcp add ados-lan -- npx -y @altnautica/ados-mcp --target agent <host>

The credential is scoped and revocable, and stored on the backend only as a hash. Revoke it in the tab to cut a server off instantly.

Then ask your client for fleet status. See docs/ for the full connect recipes and the tool catalog.

Transports

• Streamable HTTP (POST /mcp, single endpoint, SSE upgrade for streams), the primary networked transport.
• stdio, the local one-liner for Claude Code and Desktop.
• Unix socket (/run/ados/mcp.sock), on-box in agent-mode, where local presence is the credential.

Safety model

• A token carries scopes (read, safe_write, admin, flight, destructive, secret_read). Flight and destructive are off by default.
• Every write is checked against the token scope, a per-tool safety class, and, for flight or destructive actions, a typed confirmation and an operator-present signal.
• Annotations (readOnlyHint, destructiveHint) are advertised honestly but are hints, never the enforcement point. The server enforces.
• Every call produces one redacted audit event.

Develop

pnpm install
pnpm typecheck
pnpm lint
pnpm test
pnpm build

The connector is a thin shim, so it holds no telemetry buffers and runs no inference. It calls the platform's existing interfaces and presents them through MCP.

Contributing

Issues and pull requests are welcome. The repository is developed in the open under the MIT license. Please keep contributions technical and free of any private or third-party content.

License

MIT. See LICENSE. Built for civilian use.