Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/alifanov/ai-garage-launchWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/alifanov/ai-garage-launch/validate)<a href="https://agentmods.dev/commands/alifanov/ai-garage-launch/validate"><img src="https://agentmods.dev/badge/commands/alifanov/ai-garage-launch/validate.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00028 | $0.00498 |
| Opus 5 | $0.00014 | $0.00249 |
| Sonnet 5 | $0.00006 | $0.00100 |
| Haiku 4.5 | $0.00003 | $0.00050 |
Grade A, and why
validate scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 7d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
/garage:validate — 5 вопросов-фильтров
Предусловия
Прочитай docs/STATE.md и docs/01-ideas.md. Если их нет — остановись и скажи: сначала запусти /garage:start.
Что сделать
Возьми донора (из аргумента или фаворита в docs/01-ideas.md) и ответь на 5 вопросов. Используй веб-поиск и скиллы .archived:similarweb (трафик), dataforseo-keyword-research (спрос по ключам), Indie Hackers (доход).
- Есть ли трафик у конкурента? — проверь через SimilarWeb (порядок величины визитов/мес).
- Зарабатывает ли он? — данные с Indie Hackers / открытых источников (MRR, число клиентов).
- Какова его ОДНА главная фича? — не весь продукт, а то, за что реально платят.
- Могу ли я сделать это сам? — честная оценка сложности для соло + AI за часы/дни, а не месяцы.
- Какой канал трафика я могу использовать? — SEO / Reddit / Product Hunt / Google Ads / партнёрки — что реально для тебя.
Вердикт: GO только если на все 5 есть внятный ответ. Иначе — NO-GO, вернись к /garage:start и возьми другого донора.
Артефакт
Запиши в docs/02-validation.md: донор, ответы по 5 пунктам с цифрами/ссылками, вердикт. При GO — обнови docs/STATE.md: поле Донор = выбранный конкурент, отметь шаг 2 [x].
➡️ Дальше
- GO →
/garage:channel— выбрать один канал трафика (матрица аудитория×канал). - NO-GO →
/garage:start— взять другого донора.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 7d ago First seen · 32 lines · 28 tokens per session scan A ec10e93c9091
validate is a command published in the GitHub repository alifanov/ai-garage-launch (2 stars, last pushed 29d ago), licensed MIT. It adds 28 tokens to every session and 498 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
save
Save current work state for next session. Creates/updates .planning/ files (CHECKPOINT.md, STATE.md, settings.json) so Heimdall resumes with full context. NOT a rewind — saves forward progress. Run before closing a session or at any milestone.
code-review
Multi-model code review — AI models independently review code, then converge on findings.
plan-review
Multi-model plan review — AI models independently plan, then converge on the best approach.
review
Multi-model review — AI models independently review any document or general topic, then converge on findings.
maintain
Run automated maintenance — seeker finds bugs from pod logs and raises GitHub issues, fixer picks them up and creates PRs. Can run as a one-shot or scheduled via /schedule.
autonomy
Set Heimdall autonomy (1=Guided, 2=Checkpoint, 3=Full Auto) — how much the agent does before asking. Use with a number, +/- to cycle, or no argument to show current.