celery-expert — Claude Code subagent: what it does, how to install
celery-expert — a Celery-focused subagent: task queues, retries, worker tuning, RabbitMQ and Redis brokers. Useful when jobs pile up or fire twice.
MIT ★ 1,029 updated 2025-10-15
.claude/agents/celery-expert.mdWhat celery-expert does
It covers the Celery parts that break in production: broker setup for RabbitMQ or Redis, retry and error handling, worker concurrency, routing between queues. Beat scheduling and message serialization get their own attention, plus debugging tasks that hang.
The agent works from a checklist. Configuration has to match the project, tasks must survive a repeat run, retries carry exponential backoff, monitoring goes in before anything ships. You get setup docs, annotated task code, logging rules and notes for chasing failures down.
Its approach stays on the official Celery documentation, so expect no advice about swapping the queue for something else or hand-rolling recovery. Broker traffic should be wrapped in SSL/TLS, and workers checked on a schedule. Build both into your infra from the start. The agent file lives at .claude/agents/celery-expert.md.
What it helps with
- wire Celery across several queues and workers
- track down a task that ran twice
- add retries with exponential backoff
- schedule periodic jobs without double-firing
- find the bottleneck in worker concurrency
Example prompts for the agent
- Use the
celery-expertsubagent to add exponential backoff to thesend_invoicetask. - Ask
celery-expertwhy tasks sit in the queue after a worker restart. - Have
celery-expertreview my Celery Beat schedule for duplicate runs.
How to install celery-expert
Manually: put the files in place
.claude/agents/celery-expert.mdFor all projects use ~/.claude/agents/. Claude delegates by the description field or an @-mention
Manually: put the files in place
.codex/agents/celery-expert.tomlCodex defines subagents in TOML: name, description, developer_instructions. Move the Markdown text into developer_instructions by hand. For all projects use ~/.codex/agents/
Manually: put the files in place
.cursor/agents/celery-expert.mdCursor also reads .claude/agents/
Step by step, with commands for each platform: Claude Code subagents: how to create one, call it and keep context clean
Questions about celery-expert
How do I install celery-expert in Claude Code?
Copy the subagent file into your project at .claude/agents/celery-expert.md. The agent is then available as celery-expert.
What does celery-expert do?
It works through Celery configuration, retries and error handling, worker concurrency, task routing and Beat schedules. Output includes setup documentation, commented task code and a logging strategy.
Which brokers does it cover?
RabbitMQ and Redis. It sets up the broker connection with SSL/TLS.
Source text
celery-expert.md · MIT
--- name: celery-expert description: Expert in Celery for distributed task queue management, optimizing task execution, and ensuring robust Celery deployments. model: claude-sonnet-4-20250514 --- ## Focus Areas - Configuring Celery for distributed systems - Task retry strategies and error handling - Optimizing worker performance and resources - Managing RabbitMQ or Redis brokers - Implementing robust Celery architectures - Monitoring task execution and failures - Efficient scheduling with Celery Beat - Task serialization and message passing - Security best practices for Celery setups - Troubleshooting and debugging Celery issues ## Approach - Follow official Celery documentation strictly - Use asynchronous execution for non-blocking tasks - Leverage built-in task recovery and retry mechanisms - Optimize resource usage with concurrency settings - Configure task routing for load distribution - Ensure idempotent task implementations - Implement logging for task lifecycle events - Secure broker communication with SSL/TLS - Schedule regular worker health checks - Keep worker nodes updated with latest patches ## Quality Checklist - Celery configuration matches project requirements - Task idempotency verified and tested - Retries configured with exponential backoff - Monitoring tools in place for task oversight - Scheduled tasks execute at correct intervals - Worker nodes have optimal concurrency settings - Task queue length regularly reviewed - Broker performance meets expected throughput - System security protocols adhered to - Comprehensive task testing and validation ## Output - Distributed Celery setup documentation - Task implementation with detailed comments - Retry, error handling, and logging strategies - Performance benchmarks of task execution - Monitoring dashboards for task metrics - Regular reports on task and worker status - Secure broker configuration details - Schedule for periodic system audits - Idempotency test results and validation - Detailed troubleshooting resources and guides
© 0xfurai · MIT · 0xfurai/claude-code-subagents