celery_root 0.2.0

Command & Control for Celery Workers

Celery Root

Docs: https://docs.celeryroot.eu

Celery Root is a control plane for Celery. It provides a Django-based UI, an event listener/collector, and helper utilities for inspecting tasks, workers, queues, and beat schedules. The Python package and distribution remain celery_root for compatibility.

Features

• Task list with filtering, sorting, and detail views (args/kwargs/result/traceback).
• Task relation graph visualization (chains, groups, chords, maps).
• Worker fleet overview and per-worker drill-down.
• Broker queue inspection and purge actions.
• Beat schedule overview and editor.
• Pluggable storage (SQLite by default).

Quickstart (demo)

Requirements: Python >= 3.13, uv, and Docker (for the demo broker/redis).

export CELERY_ROOT_WORKERS="your_app.celery:app,another_app.celery:app"

Start the supervisor + UI (standalone):

celery_root -A your_app.celery:app

Or run as a Celery subcommand:

celery -A your_app.celery:app celery_root

By default the UI binds to 127.0.0.1:8000.

Demo stack

Requirements: Python >= 3.10, uv, and Docker (for the demo broker/redis).

make demo-infra
make demo-worker-math
make demo-worker-text
make demo-root

Then open http://127.0.0.1:8000.

To enqueue demo tasks:

make demo-tasks

Installation (repo)

Celery Root is currently built and run from this repository.

make install

This runs:

• uv sync --all-extras --dev --frozen
• uv run pre-commit install
• npm --prefix frontend/graph-ui install

Build the frontend assets:

celery_root -A demo.worker_math:app

Via Celery:

celery -A demo.worker_math:app celery_root

Configuration

Configuration is explicit via Pydantic models. Components are enabled when their config is provided (set to None to disable).

from pathlib import Path

from celery_root import (
    BeatConfig,
    CeleryRootConfig,
    DatabaseConfigSqlite,
    FrontendConfig,
    LoggingConfigFile,
    OpenTelemetryConfig,
    PrometheusConfig,
)

config = CeleryRootConfig(
    logging=LoggingConfigFile(log_dir=Path("./logs")),
    database=DatabaseConfigSqlite(db_path=Path("./celery_root.db")),
    beat=BeatConfig(schedule_path=Path("./celerybeat-schedule")),
    prometheus=PrometheusConfig(port=8001, prometheus_path="/metrics"),
    open_telemetry=OpenTelemetryConfig(endpoint="http://localhost:4317"),
    frontend=FrontendConfig(host="127.0.0.1", port=5555),
)

The web UI reads worker import paths from CELERY_ROOT_WORKERS (comma-separated). If you need to override settings before Django settings load:

from celery_root.config import set_settings

set_settings(config)

Library usage

Start the supervisor from Python:

from celery_root import CeleryRoot

root = CeleryRoot("your_app.celery:app")
root.run()

MCP server (AI tools)

Celery Root ships with an optional MCP server that exposes read-only tools over HTTP. It is designed for MCP clients (Codex CLI, Claude Code, etc.) to inspect the Celery Root store safely without write access.

Configuration:

• CELERY_ROOT_MCP_ENABLED: Enable the MCP server (1/true).
• CELERY_ROOT_MCP_HOST: Host interface (default: 127.0.0.1).
• CELERY_ROOT_MCP_PORT: Port (default: 9100).
• CELERY_ROOT_MCP_PATH: Base path (default: /mcp/).
• CELERY_ROOT_MCP_AUTH_KEY: Required auth token for clients.
• CELERY_ROOT_MCP_READONLY_DB_URL: Deprecated (RPC-based access replaces direct DB reads).

Example:

export CELERY_ROOT_MCP_ENABLED=1
export CELERY_ROOT_MCP_AUTH_KEY="your-secret-token"

Start the supervisor (or MCP server) and open the Settings page to copy client snippets.

Development

Run checks locally:

uv run precommit
uv run mypy
uv run pytest

Project structure

• celery_root/components/: optional components (web, metrics, beat).
• celery_root/core/: engine + DB + logging internals.
• demo/: demo workers and task scripts.
• tests/: unit and integration tests.