Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add sina-heidariaan/tornhill/plugin install tornhillWrote 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/sina-heidariaan/tornhill/audit)<a href="https://agentmods.dev/skills/sina-heidariaan/tornhill/audit"><img src="https://agentmods.dev/badge/skills/sina-heidariaan/tornhill/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/sina-heidariaan/tornhill/audit"><img src="https://agentmods.dev/badge/skills/sina-heidariaan/tornhill/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 | $0.00186 | $0.02637 |
| Opus 5 | $0.00093 | $0.01319 |
| Sonnet 5 | $0.00037 | $0.00527 |
| Haiku 4.5 | $0.00019 | $0.00264 |
Grade A, and why
audit scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 9d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 210 lines — stays where its author put it; the contents beside it link to each section on GitHub.
audit Skill
Produces a grounded compliance/security/legal/cost review. A generic "you should
use HTTPS and validate input" checklist is the failure mode this skill exists to
avoid. Every finding must point at a real path:line AND cite a real standard
clause — or be labelled unmapped.
Companion to
/tornhill(architecture). Same ethos: deterministic miners → cite-able candidates → reasoned findings → markdown source + HTML twin.
⛔ Grounding contract (overrides everything)
Two anchors per finding — miss either and you drop or downgrade it:
- Code anchor. Every finding cites a real
path:line(fromroutes.json,signals.json, or a handler trace). No "the app probably…". Candidates fromtornhill-scan-signals.pyare candidates — confirm each against the real code before it becomes a finding; drop the ones that aren't real. - Standard anchor. A finding may attach a standard ONLY via a
standard_id + clause_idthat resolves in the bundled catalog (${CLAUDE_PLUGIN_ROOT}/skills/audit/standards/). If nothing in the catalog fits, either downgrade tounmapped(a real code issue with no standard attribution) or verify the clause live against the standard'sauthority_url(WebFetch) and stampverified_at+ URL. Never invent a clause number or a regulation. See${CLAUDE_PLUGIN_ROOT}/docs/audit-standards.md.
Surface catalog_version in the report so a reader sees how fresh the mapping is.
Pipeline
1. Establish the profile (domain + jurisdiction) — auto-detect, then CONFIRM
The applicable standards depend entirely on what the app is and where it operates. Get this right first.
- Auto-detect signals:
- domain — dependencies (
stripe/braintree/square→ payments + card data;fhir/hl7/smart-on-fhir→ healthcare;plaid/ISO 20022 libs → financial; ecommerce carts), directory/keyword hits. - jurisdiction/region — i18n locale dirs, currency/locale config, address/phone formats, cloud-region config, an existing privacy policy, company region.
- domain — dependencies (
- Propose a profile
{domains, jurisdictions, handles: [card_data, health_data, pii], has_ui}and ask the user to confirm or edit it. Accept explicit overrides. If a committedtornhill/audit/profile.ymlexists, offer to reuse it. - Persist the confirmed profile to
tornhill/audit/profile.yml, stamped withcatalog_version+derived-from, so re-runs are reproducible.
What ships with it
9 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.
- 9d ago First seen · 210 lines · 186 tokens per session scan A 41774c01811d
audit is a skill published in the GitHub repository sina-heidariaan/tornhill (1 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 186 tokens to every session and 2,637 once invoked, about $0.0009 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 skills, from other repositories
qector-licensing
QECTOR licensing, tier enforcement, and key activation. Covers the three license tiers (Community d<=7, Pro d<=19, Enterprise d<=63), the Ed25519-signed token formats (QECT-PRO-, QECT-ENT-, v2), the resolution order (QECTORLICENSEKEY, QECTORLICENSEFILE, /.qector/license.key), the QECTORENFORCE / QECTORSILENT /…
beopsuny
A Korean-law research and contract-review assistant for in-house lawyers and compliance teams. It checks current primary sources, such as official laws and court decisions, and shows how reliable each source is.
cognitive-interface-audit
Comprehensive cognitive interface audit with two modes and a single tier. Planning mode designs task models, user expertise mapping, and error tolerance strategy. Audit mode evaluates mental model alignment (Norman's Gulfs), consistency (Gestalt), error tolerance (Wood 7-layer defense), cognitive load (NASA-TLX…
optimization-audit
Comprehensive optimization audit with two modes and a single tier. Planning mode designs performance strategy, capacity planning, and scaling architecture. Audit mode scans code and infrastructure for performance anti-patterns, inefficient algorithms, N+1 queries, missing caching, concurrency issues, and resource…
architecture-audit
Comprehensive architecture audit with two modes and a single tier (beta). Planning mode evaluates architectural patterns, bounded context decomposition, dependency direction, and SOLID compliance for new systems. Audit mode scans existing codebases for architectural drift, coupling violations, domain model quality…
documentation-audit
Comprehensive documentation audit with two modes and a single tier. Planning mode designs documentation strategy, content type maps, learning paths, and scaffolding architecture. Audit mode evaluates linguistic precision (Strunk & White, Google/Microsoft style guides), structural taxonomy (Diataxis quadrant…