← ClaudeAtlas

hirelisted

Hiring — design per-role model assignments and an operating profile from available models, budget and project character; applied on the budget answer (from /kickoff, /assess or this command) and shown as a roster — the CEO changes a line by saying so. Right after /kickoff or /assess, when the budget changes, or when the team is too slow or expensive
jaewhi-park/GARAGISTE · ★ 1 · AI & Automation · score 77
Install: claude install-skill jaewhi-park/GARAGISTE
Hire the team (assign a model and effort per role). Note: $ARGUMENTS 1. Candidates: the default candidates are the aliases opus / sonnet / haiku (they resolve to the plan's latest models). Include other model IDs (API, Bedrock, Vertex) only if the CEO names them. Plan and usage caps cannot be read from the CLI — ask, do not guess. 2. Conditions — skipped when the caller passed them; otherwise one AskUserQuestion call: budget tier (unlimited = API/in-house / high = Max 20x / medium = Max 5x / low = Pro), project character (docs/BRIEF.md via team-planner's report or the brainstorm session, docs/CHARTER.md and ASSESSMENT.md — a legacy rebuild weights critic and reviewer; a prototype weights speed), whether parallel development is planned, and any role to make deliberately strong or cheap. 3. Proposal: a role × model × effort table with a one-sentence reason per row and a note of where tokens concentrate. Principle: strong model and high effort where judgment happens (planner, critic, reviewer); verifier on haiku·low; implementer by budget. Also propose an operating profile: default review lenses (2 or 4), parallelism (none / session-level / /parallel), plan-size target (default 4 logic steps — the critic reviews plans over it; bigger work is split by tryable outcome), per-step verifier (off by default — the implementer's red → green report carries each step and team-verifier gives the full verdict at the end of /build; on when the project wants an independent quick run per step