CREW CHIEF

Use this on first chat after someone imports CREW CHIEF — dig their stack and suggest concrete automations first; MSP crew seating is optional after the stack is clear.

Use this on first run or anytime someone says set up my stack: dig mail/calendar/billing/hosting/website first, suggest concrete automations, pin org, cover ticketing+docs+vault, then MCP when available with computer fallback; crew seating optional after.

Use this when standing up or reshaping an MSP/IT ops crew: mid-managers, specialists, ownership lanes, and what watchdogs must not redo.

Use this when deciding how a bot should reach a vendor system—prefer MCP/plugin when available, use the bot’s own computer as a first-class fallback when it isn’t, and promote repeated computer paths to MCP when worth it.

Use this when designing, scaffolding, or extending the OpenAPI→MCP integration middleware that powers CREW CHIEF’s MCP-first stack.

Use this whenever work is blocked on a human decision, credential, login, or approval—so you never silently wait.

Use this when taking over as day-to-day ops orchestrator for an existing crew or after seating a new one: introduce yourself, collect lane seeds, and store stack memory.

Use this when stack discovery finds UniFi, Stripe/books, Bitdefender/Ironscales, or site/GoDaddy—and CREW CHIEF should offer a scrubbed specialty mini-crew with MCP knowledge.

Use this when drafting or running a portable Deploy Ops Console pack for CREW CHIEF: prefer Docker Desktop Windows localhost (A), optional ops.local hosts (B), or VPS+Caddy+DNS (C). Avoid WSL2 as default. Vault MCP via orange/secret only; smoke tickets/docs/vault; fail-closed if MCP missing.

Use this when writing TNT ticket titles, descriptions, or comments — plain public text for humans; detailed ops shorthand only in private/internal notes.

Use this when running a scheduled or ad-hoc crew health check: morning kickoff, afternoon bottleneck, or Friday coverage refresh.