Installs the golden-path harness into this repo — greenfield or grafted onto existing code — and proves the gate bites, red fails, green passes, before more code lands.
Free & open · no signup · read-only — it ends by asking · nothing leaves your machine
You are working inside this repo — empty or already full of code. Mission: install the goal-prompts golden-path harness and prove it bites — red fails, green passes — before more product code lands.
This is a Build brief: it writes files, but only after the operator approves the plan at the Phase 2 gate.
Phase 2 sweeps the codebase through every one of these, citing file and line for each finding.
a documented go, or an explicit operator ask? A hunch is not a mandate.
can this product's quality be checked by commands: tests, schema validation, golden files, rubric-scored evals with a numeric floor? If quality is pure taste, say so — this harness is the wrong tool.
greenfield (empty repo: copy the whole template) or graft (existing code: install the harness layer plus the SPEC.md and DECISIONS.md skeletons, clobbering nothing)?
what test, lint, and build commands already run here? Propose the one-line wiring of each into scripts/check; the operator ratifies those exact lines at this gate, because the harness is operator-owned.
the template default is Python 3 stdlib + unittest, zero dependencies. Any deviation needs a DECISIONS.md entry; name it now.
what only a human can do: push to GitHub, make the check workflow a required status check, point CODEOWNERS at themselves.
One structured report at the repo root — or in reports/, if you keep one — the same shape every time, ready for a teammate — or the optional Studio — to act on.
Every Goal Prompt follows the same four steps, so results are consistent and repeatable — no matter which one you run.
Find the mandate.
Score the installation through these lenses.
Copy the agreed files into this repo.
Create SCAFFOLD.md at repo root.
Copy it and paste it into your agent inside the repo you want checked.
Install the goal plugin once — two commands — then just type /goal:scaffold-the-rails.
/plugin marketplace add GhostlyGawd/goal-prompts/plugin install goal@goal-promptsOr install only this Goal Prompt as /goal-scaffold-the-rails:
curl -fsSL https://goal-prompts.vercel.app/install | BRIEF=141 shLet an agent fetch it mid-conversation, or pull the raw Goal Prompt by URL.
https://goal-prompts.vercel.app/raw/141.mdNothing hidden — this is the whole Goal Prompt, verbatim. Read it in a minute, edit it, or copy it as-is.
# Goal: Scaffold the Rails You are working inside this repo — empty or already full of code. Mission: install the goal-prompts golden-path harness and prove it bites — red fails, green passes — before more product code lands. This is a Build brief: it writes files, but only after the operator approves the plan at the Phase 2 gate. ## Phase 1 — Locate the mandate and the template - Find the mandate: a `VERDICT.md` go ruling (at this root or in `reports/`), or the operator's explicit ask to put this repo on rails. Quote it. - Locate the template: the `template/` directory of the goal-prompts repo — a local checkout, or `git clone --depth 1 https://github.com/GhostlyGawd/goal-prompts`. - Read `template/README.md` end to end; it is the contract for what you are about to install. ## Phase 2 — Audit readiness, then ask Score the installation through these lenses; cite evidence: 1. **Mandate** — a documented go, or an explicit operator ask? A hunch is not a mandate. 2. **Mechanical verifiability** — can this product's quality be checked by commands: tests, schema validation, golden files, rubric-scored evals with a numeric floor? If quality is pure taste, say so — this harness is the wrong tool. 3. **Mode** — greenfield (empty repo: copy the whole template) or graft (existing code: install the harness layer plus the SPEC.md and DECISIONS.md skeletons, clobbering nothing)? 4. **Existing checks** — what test, lint, and build commands already run here? Propose the one-line wiring of each into `scripts/check`; the operator ratifies those exact lines at this gate, because the harness is operator-owned. 5. **Stack fit** — the template default is Python 3 stdlib + unittest, zero dependencies. Any deviation needs a DECISIONS.md entry; name it now. 6. **Operator duties** — what only a human can do: push to GitHub, make the `check` workflow a required status check, point CODEOWNERS at themselves. Then stop. Report only — end by asking whether to install as planned, adjust, or abort. Nothing is written until the operator answers. ## Phase 3 — Install and prove 1. Copy the agreed files into this repo; `git config core.hooksPath .githooks`. Never overwrite an existing file unless the plan listed it. 2. Run `scripts/check` — must pass green before any new product code. 3. Run `scripts/check --prove-red` — plants a failing canary and asserts the gate goes red; a gate that cannot fail is not a gate. Capture both outputs. 4. Fill ADR-0001 in DECISIONS.md (product, date, mode, wiring, deviations) and commit as `scaffold: harness from goal-prompts template`. ## Phase 4 — Report Create `SCAFFOLD.md` at repo root: 1. **What was installed** — files added; in graft mode, the existing commands now wired into the gate 2. **Harness proof** — the green run and the prove-red run, verbatim tails 3. **Operator TODOs** — the Phase 2 duties still open, as checkboxes 4. **Next** — run 142 · Spec the Product to write SPEC.md before building Start the report with today's date. If `SCAFFOLD.md` already exists from a previous run, this repo is already on rails — verify the harness still bites and lead with what changed since. ## Rules - Nothing is written before the Phase 2 gate; the operator's answer is the scope - Graft mode clobbers nothing: existing files are never overwritten, and existing tests keep running — now inside the gate - The installed harness layer (`scripts/`, `.githooks/`, `.github/`, `.claude/`, `tests/harness/`) is operator-owned from the moment it lands - No go verdict and no operator ask to put this repo on rails? Say so in a one-paragraph null report and stop — a null result is a valid finding. - If a `reports/` directory exists at the repo root, write the report there instead of the root. - Before asking, present the top findings as a ranked list in plain words - Report only — end by asking at the Phase 2 gate before creating anything
scripts/, .githooks/, .github/, .claude/, tests/harness/) is operator-owned from the moment it landsreports/ directory exists at the repo root, write the report there instead of the root.Curated neighbors — briefs that answer the adjacent question, worth running in the same session.
The decider. Reads every venture report at this root, scores the candidate against bars set before scoring, and rules go, pivot, or kill — reasoning shown.
Turns venture research and operator intent into SPEC.md — a contract where every acceptance criterion carries a command a machine can run.
Turns venture research and operator intent into SPEC.md — a contract where every acceptance criterion carries a command a machine can run.
The build loop. Takes the open acceptance criteria, asks the scope, then lands each one as a failing test, the least code to pass it, and a verified commit.
The adversarial go/no-go. Re-runs every check fresh, sabotages code to prove the tests bite, walks the revenue path end to end, and rules ship or hold.
Runs inside these playbooks — curated sequences you can launch with one paste: