Update to Prefect

{
  "name": "workflows",
  "description": "Inspect and troubleshoot Prefect workflows, deployments, work pools, and releases with Prefect MCP tools. Use for Prefect operational questions, failure diagnosis, documentation lookup, and safe guidance when a requested mutation requires a separately configured Prefect CLI.",
  "included_files": [],
  "skill_md_contents": "---\nname: workflows\ndescription: Inspect and troubleshoot Prefect workflows, deployments, work pools, and releases with Prefect MCP tools. Use for Prefect operational questions, failure diagnosis, documentation lookup, and safe guidance when a requested mutation requires a separately configured Prefect CLI.\n---\n\n# Prefect Workflows\n\nUse the Prefect MCP tools as the canonical source for Prefect state, documentation,\nand release notes. They return structured data with complete identifiers and are\nread-only unless a tool explicitly says otherwise.\n\n## Choose a Workspace\n\nThe hosted MCP server authenticates with Prefect Cloud OAuth. If a tool needs a\n`workspace_id` and the user has not selected one:\n\n1. Call `list_authorized_workspaces`.\n2. Use the only workspace automatically when exactly one is available.\n3. When several are available, choose from explicit conversation context or ask\n   the user which workspace they mean.\n4. Pass the selected `workspace_id` consistently to subsequent tool calls.\n\nNever guess a workspace ID or copy one from an unrelated result.\n\n## Diagnose Operational Problems\n\nStart broad and narrow only as needed:\n\n1. Use `get_dashboard` for an overview of recent failures, lateness, deployments,\n   and work-pool health.\n2. Use `get_flow_runs` to inspect the affected run and preserve its full UUID.\n3. Use `get_deployments`, `get_work_pools`, `get_task_runs`, or `read_events` for\n   the specific layer implicated by the run.\n4. Explain the evidence, distinguish the observed failure from likely causes,\n   and recommend the smallest next check or action.\n\nDo not repeatedly fetch broad lists when a specific ID is already known.\n\n## Documentation and Release Notes\n\nUse the Prefect documentation tools for current CLI, SDK, deployment, and release\ninformation. For release questions, search for the current structured release\nnotes and report the exact version and release date returned by the source.\n\nDo not answer version-sensitive questions from memory when the documentation tools\nare available.\n\n## Mutations and the Prefect CLI\n\nThe hosted Prefect MCP surface is read-only. Do not claim that it triggered,\ncancelled, created, updated, or deleted anything unless an exposed tool actually\nperformed that operation.\n\nA local Prefect CLI is a separate, advanced path. Use it only when all of these are\ntrue:\n\n- the current client can execute local shell commands;\n- `prefect version` succeeds;\n- the local Prefect profile is already authenticated to the intended workspace;\n- the user explicitly requested the mutation.\n\nOtherwise, explain that the connected MCP tools are read-only and provide the\nexact CLI command for the user to run without pretending it was executed.\n\nWhen a configured local CLI is appropriate:\n\n- prefer complete UUIDs obtained from MCP results;\n- use `prefect --no-prompt` for commands that would otherwise prompt;\n- prefer `prefect api` or `-o json` when machine-readable output is needed;\n- never use a truncated identifier from a rendered table.\n\nCommon commands:\n\n```bash\nprefect deployment run 'flow-name/deployment-name'\nprefect --no-prompt flow-run cancel <flow-run-uuid>\nprefect automation create --from-file automation.yaml\n```\n"
}