/g The Contract Codex ← 返回动态 Catalog

Build Design Tool From Scratch

Run a long-horizon Codex task to build a design tool with milestone verification.

难度:intermediate 分类:greenfield-build 来源:OpenAI Codex blog

完整 Prompt(可直接复制)

/goal
GOAL:
Complete Build Design Tool From Scratch for a greenfield implementation task: Run a long-horizon Codex task to build a design tool with milestone verification.

CONTEXT:
- Before editing, read the nearest AGENTS.md/CLAUDE.md, current issue or PLAN.md, and any failing logs already in the repo.
- Inspect product requirements, milestones, tests, and verification notes.
- Establish a baseline by running or locating evidence for: `tests, lint, and typecheck per milestone`.

CONSTRAINTS:
- Keep the scope limited to this goal; do not expand into unrelated cleanup.
- Do not weaken tests, delete assertions, or mask errors to make verification pass.
- Respect the repository's AGENTS.md/CLAUDE.md instructions and existing patterns.
- Keep implementation tied to the stated product requirements.
- Do not ship a demo-only path without tests or runnable verification.

DONE WHEN:
- The implementation or documentation directly satisfies: Run a long-horizon Codex task to build a design tool with milestone verification.
- The verification command or evidence path succeeds: `tests, lint, and typecheck per milestone`.
- The final diff is scoped to the relevant files and has no unrelated formatting churn.

VERIFY:
- Run `tests, lint, and typecheck per milestone` or the closest repo-local equivalent if the exact command is not available.
- Capture before/after evidence for the behavior, metric, report, or artifact involved.
- If verification cannot run locally, stop and report the missing dependency instead of guessing success.

OUTPUT:
- Summarize changed files, key decisions, verification output, and remaining risks.
- Include any follow-up that is required for production rollout or human review.

STOP RULES:
- Pause if secrets, production access, stakeholder decisions, or destructive data operations are required.
- Pause after three failed fix attempts on the same symptom and challenge the root-cause hypothesis.
- Do not mark the goal complete until the current repository state has been audited against DONE WHEN.

来源与证据

原始来源: OpenAI Codex blog

证据摘要: verification steps; source: OpenAI Codex blog; type: official-workflow; verification: tests, lint, and typecheck per milestone