Getting Started

Installation

Homebrew (macOS / Linux)

brew install bradleydwyer/tap/tasked

Build from source

git clone https://github.com/bradleydwyer/tasked
cd tasked && cargo build --release

The binary is at target/release/tasked-server.

Verify

tasked --help

Your first flow

A flow is a JSON file defining tasks and their dependencies. Create a file called flow.json:

{
  "tasks": [
    {
      "id": "hello",
      "executor": "shell",
      "config": { "command": "echo 'Hello from Tasked!'" }
    },
    {
      "id": "world",
      "executor": "shell",
      "config": { "command": "echo 'DAG execution works!'" },
      "depends_on": ["hello"]
    }
  ]
}

This defines two tasks. world depends on hello, so the engine runs hello first, then world.

Run it

tasked run flow.json

You'll see:

▸ Flow f_7k2m submitted (2 tasks)
✓ [hello]  succeeded  0.1s
✓ [world]  succeeded  0.1s

✓ Flow complete  2/2 tasks succeeded (0.3s)

That's it. Tasked validated the DAG, resolved execution order, ran each task, and reported results.

Server mode

For production use, run Tasked as an HTTP server:

# Start the server
tasked-server serve --port 8080

# Create a queue
curl -X POST http://localhost:8080/api/v1/queues \
  -H "Content-Type: application/json" \
  -d '{"id": "default"}'

# Submit a flow
curl -X POST http://localhost:8080/api/v1/queues/default/flows \
  -H "Content-Type: application/json" \
  -d @flow.json

# Check flow status
curl http://localhost:8080/api/v1/flows/{flow_id}

See the API Reference for all endpoints.

MCP mode

Give AI agents access to durable task execution via the Model Context Protocol:

tasked-server mcp --data-dir tasked-data

Agents can submit flows, check status, and retrieve task outputs through MCP tools. See the MCP guide.

Next steps

• Concepts — understand DAGs, task states, and variable substitution
• Executors — shell commands, HTTP requests, callbacks
• Flows — full flow definition reference
• Queues — concurrency, rate limiting, retry policies