build-agentlisted

# Build Agent Also fires on 'build-agent' or any request to stand up a Claude Managed Agent in a client's Anthropic account, scoped to a specific repo. ## What this skill does Takes a **build-sheet** (the contract written by `/imagine-agent`, or one this skill produces itself) and stands up a live **Claude Managed Agent** in the client's Anthropic account: stage payloads → launch the agent → run a graded outcome → iterate → schedule it on cron. The agent is scoped to a repo (cloned into the session) and keeps running in the client's Console after this session ends. This is the technical half of the pair. The non-technical client interview lives in `/imagine-agent`. ## When to load sub-files - Need the CMA API (auth header, endpoints, payload shapes, launch order, limits) → read [references/cma-primitives.md](references/cma-primitives.md). **Always read this before any curl** — the auth header (`x-api-key`, not Bearer) is the #1 footgun. - Need the build-sheet schema / validation checklist → read [references/build-sheet.md](references/build-sheet.md). ## Preflight — do this first, every time 1. **Read `references/cma-primitives.md`.** Don't curl from memory. 2. **Smoke-test the key + CMA access** (substitute the client's key — see Account model below): ```bash curl -s -o /dev/null -w "auth %{http_code}\n" https://api.anthropic.com/v1/models \ -H "x-api-key: $ANTHROPIC_API_KEY" -H "anthropic-version: 2023-06-01" curl -s -o /dev/null -w "cma %{http_cod