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 skills add haru0416-dev/quaere --skill quaere-auditgit clone --depth 1 https://github.com/haru0416-dev/quaereWrote 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/skills/haru0416-dev/quaere/quaere-audit)<a href="https://agentmods.dev/skills/haru0416-dev/quaere/quaere-audit"><img src="https://agentmods.dev/badge/skills/haru0416-dev/quaere/quaere-audit/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/haru0416-dev/quaere/quaere-audit"><img src="https://agentmods.dev/badge/skills/haru0416-dev/quaere/quaere-audit.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00146 | $0.04132 |
| Opus 5.5 | $0.00058 | $0.01653 |
| Sonnet 5.5 | $0.00029 | $0.00826 |
| Haiku 4.5 | $0.00015 | $0.00413 |
Grade A, and why
quaere-audit 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 1mo 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.
How it starts
The opening of the file, as written. The whole thing — 270 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Security Audit Loop
Iron Law
No security finding is confirmed without a falsifiable property, a reachable attacker-controlled path, a missing or failed guard, concrete impact, and a disconfirming false-positive pass.
This is the security version of evidence-gating. Dangerous APIs, scary strings, stale advisories, and pattern matches are leads, not findings. A finding becomes actionable only when the audit can show why the property should hold, where an attacker crosses a boundary, which sink/state/guard fails, what impact follows under the threat model, and what counter-evidence was checked. If any link is missing, classify it as potential, rejected, or inconclusive.
When to use
- Vulnerability discovery, security code review, bug bounty preparation, exploitability triage, or protocol/spec-grounded auditing.
- Auth/authz, tenant isolation, injection, parser, sandbox, cryptography, consensus, payments, secrets, SSRF, deserialization, supply-chain, cloud/config, or CI/CD security work.
- Security-sensitive PRs or incidents where false positives, unsafe probes, or stale external facts would be costly.
When NOT to use
- Pure dependency lookup or CVE freshness checks with no local exploitability analysis; use
quaere-groundingfirst. - Normal implementation after a vulnerability is already confirmed and a fix is authorized; use
quaere-execution. - General maintainability review, one-line known fixes, or shallow checklist-only reviews.
- Any task whose rules of engagement are missing and probing could affect third-party or production systems; stop and ask.
Handoff triggers (which skill comes after this one)
Audit coordinates other skills; hand off when the blocking question is not security-property mapping:
- Code intent, invariants, failure modes, or downstream connections are unclear →
quaere-semantic. - Current specs / CVEs / advisories / docs / cloud behavior / framework guidance / CLI behavior / bounty rules must be anchored →
quaere-grounding. - A subtle / disputed / security-sensitive candidate finding still needs proof before being confirmed →
quaere-evidence. - Safe PoC, regression test, or authorized fix is the next step →
quaere-execution.
What ships with it
6 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 1mo ago First seen · 270 lines · 146 tokens per session scan A c4efca8f4978
quaere-audit is a skill published in the GitHub repository haru0416-dev/quaere (5 stars, last pushed 1mo ago), licensed MIT. It adds 146 tokens to every session and 4,132 once invoked, about $0.0006 per session on Opus 5.5. 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 skills, from other repositories
task-workflow
Shared implementation workflow for task-based skills, with profile-check sites wrapped in Jinja conditionals.
task-workflow
Shared implementation workflow for task-based skills, with profile-check sites wrapped in Jinja conditionals.
task-workflow
Shared implementation workflow for task-based skills, with profile-check sites wrapped in Jinja conditionals.
aitask-pickrem-remote
Pick and implement a task in remote/non-interactive mode. All decisions from execution profile - no AskUserQuestion calls.
aitask-pickrem-remote
Pick and implement a task in remote/non-interactive mode. All decisions from execution profile - no AskUserQuestion calls.
aitask-explorechat
Chat-native explore flow for a bug report — machine-spawned by the chatlink gateway, every decision routed through the file-spool Q&A relay instead of AskUserQuestion, ending in a task-creation payload.json. Not a user task command.