Menu
Run the captain's weekly Monday-plan / Wednesday-check / Friday-review cycle
Instalar com Codex ou Claude Copie este prompt, cole no Codex, Claude ou outro assistente e deixe que ele revise a página da skill e instale para você.
Baseado na classificação ocupacional SOC
Multi-slide bilingual brand carousels — Claude writes the deck, kun renders Anthropic-styled slides at exact platform sizes, a human approves, channels receive
Draft, stage, and publish brand social posts — Claude drafts, /higgs renders, a human approves, Hermes relays
Convert a file or URL to Markdown via MarkItDown (PDF, Office, images, audio, web)
Full pipeline — idea to production (chains every stage)
Autonomous block QA — detect, adversarially verify, fix safe tiers, hand the residual to a human
Technical spec — data model, file plan, refined acceptance criteria
| name | weekly |
| description | Run the captain's weekly Monday-plan / Wednesday-check / Friday-review cycle |
| when_to_use | Use when running the captain's weekly cadence — Monday plan (allocate the 7 humans), Wednesday check (progress vs plan), Friday review (coach + canon principle) with archive to memory/weekly. Triggers on: weekly, monday plan, friday review, plan the week. |
| argument-hint | [plan|check|review] |
Run the captain's weekly review.
Arguments: $ARGUMENTS (optional: "plan" for Monday planning, "check" for Wednesday, "review" for Friday)
Default behavior (no args): Full weekly review.
Steps:
/sync — refresh the engine's view of its ecosystem before planning (due tiers only; it self-skips tiers whose engine.json → sync.<tier> stamp is fresh).claude/memory/decisions/ and /decide entries) for entries whose reviewed-by date has passed: grade the prediction against what actually happened, write the lesson as a feedback_* memory, and mark the entry reviewed. Predictions that are never graded never compound.Reference: .claude/agents/captain.md for weekly rhythm Reference: .claude/memory/team.json for team profiles