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.
Free & open · no signup · read-only — it ends by asking · nothing leaves your machine
You are working inside this repo — a product repo built against SPEC.md. Mission: the adversarial go/no-go before real users or money — verify by running, never by reading, and rule ship or hold.
Read-only pass: every experiment below is reverted before this brief ends. Your only lasting write is the report file.
Phase 2 sweeps the codebase through every one of these, citing file and line for each finding.
run scripts/check fresh, then scripts/check --prove-red; a gate that cannot go red proves nothing
run every built AC's check: command verbatim; produce a pass/fail table, no paraphrase
pick 3 built ACs, deliberately break each behavior (revert after), and confirm its test goes red; a test that survives sabotage asserts nothing
diff the harness layer (scripts/, .githooks/, .github/, .claude/, tests/harness/) against the goal-prompts template; any drift is a finding, whoever made it
run python3 evals/run.py; scores against the spec's numeric floor
walk the spec's revenue path end to end: can a stranger actually pay today, and does the money arrive somewhere the operator controls?
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.
SPEC.md, restated as the first Weekly Vitals (29) checklistEvery Goal Prompt follows the same four steps, so results are consistent and repeatable — no matter which one you run.
Read SPEC.md (the bars), BUILDLOG.md (the claims), DECISIONS.md (the promises).
The gate itself — run scripts/check fresh, then scripts/check --prove-red.
The bars come from SPEC.md and were written before this run.
Create SHIP-GATE.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:ship-gate.
/plugin marketplace add GhostlyGawd/goal-prompts/plugin install goal@goal-promptsOr install only this Goal Prompt as /goal-ship-gate:
curl -fsSL https://goal-prompts.vercel.app/install | BRIEF=144 shLet an agent fetch it mid-conversation, or pull the raw Goal Prompt by URL.
https://goal-prompts.vercel.app/raw/144.mdNothing hidden — this is the whole Goal Prompt, verbatim. Read it in a minute, edit it, or copy it as-is.
# Goal: Ship Gate You are working inside this repo — a product repo built against `SPEC.md`. Mission: the adversarial go/no-go before real users or money — verify by running, never by reading, and rule ship or hold. Read-only pass: every experiment below is reverted before this brief ends. Your only lasting write is the report file. ## Phase 1 — Load the contract - Read `SPEC.md` (the bars), `BUILDLOG.md` (the claims), `DECISIONS.md` (the promises). - Note every AC claiming `status: built` — those claims are on trial. ## Phase 2 — Verify through 6 lenses 1. **The gate itself** — run `scripts/check` fresh, then `scripts/check --prove-red`; a gate that cannot go red proves nothing 2. **AC truth** — run every built AC's `check:` command verbatim; produce a pass/fail table, no paraphrase 3. **Test honesty** — pick 3 built ACs, deliberately break each behavior (revert after), and confirm its test goes red; a test that survives sabotage asserts nothing 4. **Harness integrity** — diff the harness layer (`scripts/`, `.githooks/`, `.github/`, `.claude/`, `tests/harness/`) against the goal-prompts template; any drift is a finding, whoever made it 5. **Eval floor** — run `python3 evals/run.py`; scores against the spec's numeric floor 6. **The first dollar** — walk the spec's revenue path end to end: can a stranger actually pay today, and does the money arrive somewhere the operator controls? ## Phase 3 — Rule - The bars come from `SPEC.md` and were written before this run; never adjust them to fit the results. - Any hard fail — a red AC, a sabotage-proof test, harness drift, a broken payment path — is a hold, named after its blocker. No averaging. ## Phase 4 — Report Create `SHIP-GATE.md` at repo root: 1. **Scorecard** — lens · result · evidence, as command-output tails 2. **The ruling** — ship or hold, in one plain paragraph 3. **Blockers** — if hold: finding · which brief fixes it (143 for unmet ACs, 142 for spec gaps, 47 for report findings) 4. **Watch after ship** — the kill criteria from `SPEC.md`, restated as the first Weekly Vitals (29) checklist Start the report with today's date. If `SHIP-GATE.md` already exists from a previous run, read it first and lead with what changed — especially blockers that were open and are now closed. ## Rules - Verify by executing; a gate that trusts reports is not a gate - Sabotage is always reverted: `git status` must be clean when this brief ends - No `SPEC.md`, or no implementation to judge at this root? Say so in a one-paragraph null report and stop. - 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 whether to ship, or which blocker to send back first
git status must be clean when this brief endsSPEC.md, or no implementation to judge at this root? Say so in a one-paragraph null report and stop.reports/ 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 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.
A fast weekly vitals pass — tests, build, deps, TODOs — diffed against the last run so drift shows up as trend arrows.
Turns your root reports into commits. Asks the scope — all, high-priority, or its pick — then builds them in dependency order, one verified commit each.
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.
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.
Runs inside these playbooks — curated sequences you can launch with one paste: