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.
npx agentmods add commands/vouchdev/vouch/vouch-askgit clone --depth 1 https://github.com/vouchdev/vouchWhat 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 | $0.00015 | $0.00245 |
| Opus 5 | $0.00008 | $0.00122 |
| Sonnet 5 | $0.00003 | $0.00049 |
| Haiku 4.5 | $0.00002 | $0.00024 |
Grade A, and why
vouch-ask scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
/vouch-ask
Answer "$ARGUMENTS" using only reviewed knowledge from the vouch KB. Every
statement in the answer must carry a [claim-id] or [source-id] citation.
Steps:
- Call
kb_searchwithquery: "$ARGUMENTS", thenkb_contexton the same query to assemble the working set. - If the results answer the question, write the answer citing every claim id
you relied on. Use
kb_read_pagefor typed record detail (contacts, project records, followups). - If the KB cannot support an answer, say so plainly and list the closest claims found. Do not fill gaps from your own knowledge — an uncited answer is worse than no answer.
- Only when the user explicitly asks about in-flight knowledge, list
kb_list_pendingitems — each labeledUNREVIEWED— after the cited answer, never mixed into it.
Never call kb_approve. Never restate pending proposals as facts.
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.
- 2d ago First seen · 25 lines · 15 tokens per session scan A b74107e9c5bb
vouch-ask is a command published in the GitHub repository vouchdev/vouch (92 stars, last pushed 9d ago), licensed MIT. It adds 15 tokens to every session and 245 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 2026-08-30.
Other commands, from other repositories
validate-gap
Analyze implementation gap between requirements and existing codebase.
spec-impl
Execute spec tasks using TDD methodology.
security-dialogs
Deep AI-based credential scan for dialog files.
security
Провести аудит безопасности кода.
gsd:list-phase-assumptions
Surface Claude's assumptions about a phase approach before planning.
compile-docker
Build the project using Docker. Only builds the builder stage to verify Go code compilation.