Prompt file imported from jerryboganda/oetwebapp (
.github/prompts/ultrawork.prompt.md). Fill in{{goal}}before use. Copyright stays with the author.
Run ultrawork for this goal: {{goal}}.
First read AGENTS.md, SESSION_STATE.md, TASKS.json and PROGRESS.md. Continue from SESSION_STATE.md only when its Goal matches this goal; otherwise re-goal it with pnpm run ax:init and update TASKS.json. Then perform evidence-based planning using only relevant domain docs, current code, tests, local patterns, and useful web documentation. Include acceptance criteria, touched files/contracts, risks, rejected approaches, and validation matrix. Enter autopilot implementation mode automatically; do not stop at the plan or offer manual proceed handoffs. Spawn only registered specialist agents from the Orchestrator allowlist when useful. Ask popup-style questions only when a missing user decision blocks correctness, and include recommended options. Implement, verify on GitHub Actions only (per AGENTS.md; pnpm run ship:gate and pnpm run ax:check are the only local checks), review, fix confirmed issues, record real run ids with pnpm run ax:record, update SESSION_STATE.md and TASKS.json, and continue until complete or genuinely blocked.
