Run durable agent tasks and event-driven workflows with Hatchet
Use Hatchet to queue, retry, pause, resume, observe, and scale long-running agent tasks and background workflows.
npx skills add agentskillexchange/skills --skill run-durable-agent-tasks-and-event-driven-workflows-with-hatchet
Use Hatchet when an agent workflow needs durable execution instead of a single fragile process: long-running tasks, event waits, retries, rate limits, fan-out work, scheduled runs, and observable background jobs. The operator defines tasks or workflows in application code, runs Hatchet workers, queues agent-driven jobs, and reviews task state, failures, logs, metrics, and retry behavior before follow-up automation continues. Inputs are task definitions, triggers, payloads, retry policies, concurrency limits, event waits, and worker code. Outputs are completed jobs, failure traces, durable state, logs, alerts, and metrics. The scope boundary is durable execution for agent and background workflows, not a general hosted product, SDK, task queue, framework, or cloud platform listing.
What this skill actually does
Inputs and prerequisites: Hatchet CLI, Docker for local setup, Hatchet workers, and application code in Python, TypeScript, Go, Ruby, or another supported runtime.
Setup notes: Install the Hatchet CLI from the official installer or package instructions, confirm the CLI version, start a local server with hatchet server start, then define and run workers for durable tasks, retries, event waits, schedules, and workflows using the upstream SDK examples.
Source and verification boundary: use https://docs.hatchet.run as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a Multi-Framework workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.