brainstorming
brainstorming
You MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation).
flow-task-authoringAuthor, run, debug, and locally test Prefect flows and tasks with retries, caching, states, futures, task runners, results, transactions, hooks, and concurrency.
Independent GitHub snapshot · Automated classification · Original documentation governs
Use this sub-skill when a task is about writing or debugging Prefect SDK workflow code: @flow, @task, local execution, state handling, futures, task runners, result persistence, retries, caching, logging, hooks, interactive pauses, transactions, and unit-style tests.
.submit(), .map(), state inspection, local tests, cached tasks, result persistence, retry policies, hooks, and nested subflows.python scripts/flow_task_smoke.py --help when you need a deterministic local smoke example with no network, credentials, long-running service, or original-repo dependency.Flow and Task objects created by decorators.flow.serve, flow.deploy, schedules, work pools, work queues, workers, and prefect.yaml to ../deployments-workers/SKILL.md.../cli-server-operations/SKILL.md.get_client, PrefectClient, schema models, profiles/settings internals, and temporary_settings to ../api-client-settings/SKILL.md.uv/just, generated docs, and source-tree development workflow to ../repo-development/SKILL.md.flow_fn(...) or task_fn(...) while authoring; add deployments/workers only after the workflow behavior is correct.@flow(log_prints=True) or get_run_logger() when the user expects logs to appear in Prefect run logs..submit() or .map() only inside a flow when concurrency or dependency tracking is needed; call tasks directly when sequential behavior is clearer.future.result() or PrefectFutureList.result() before returning or asserting values; use return_state=True when debugging failed states without raising immediately.persist_result=True, result_storage_key, or cache_policy=INPUTS when cache/result reuse must survive beyond one task call.scripts/flow_task_smoke.py demonstrates deterministic flow/task authoring, local HTML parsing adapted from the repo example without network, retries, cache configuration, futures, mapping, hooks, and state inspection.--mode basic, --mode futures, --mode state, --mode cache, and --mode all for focused checks.