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 trilwu/gauntlet-loop-skills --skill security-vuln-gauntletgit clone --depth 1 https://github.com/trilwu/gauntlet-loop-skillsWrote 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/trilwu/gauntlet-loop-skills/security-vuln-gauntlet)<a href="https://agentmods.dev/skills/trilwu/gauntlet-loop-skills/security-vuln-gauntlet"><img src="https://agentmods.dev/badge/skills/trilwu/gauntlet-loop-skills/security-vuln-gauntlet/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/trilwu/gauntlet-loop-skills/security-vuln-gauntlet"><img src="https://agentmods.dev/badge/skills/trilwu/gauntlet-loop-skills/security-vuln-gauntlet.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 | $0.00196 | $0.01301 |
| Opus 5 | $0.00098 | $0.00651 |
| Sonnet 5 | $0.00039 | $0.00260 |
| Haiku 4.5 | $0.00020 | $0.00130 |
Grade A, and why
security-vuln-gauntlet 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 11d 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 — 76 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Security Vuln Gauntlet
The Gauntlet Loop applied to finding real vulnerabilities: a HUNTER (builder) proposes candidates and a blind VALIDATOR (critic) demands a working PoC before anything counts. The bar is exploitability + CWE mapping + real precedent. Covers web, API, binary/memory, and cloud/IaC.
Kept separate from the core
gauntlet-loopskill because its triggers and safety requirements differ: written authorization (ROE) is mandatory and every PoC must be non-destructive. The method itself is the core loop.
SCOPE AND AUTHORIZATION — mandatory
Use only on assets you own or are authorized in writing to test. If there is no ROE, STOP and ask the user for one. Read references/safety-and-scope.md first.
Every PoC must be non-destructive: no data destruction, no lateral movement, no exfiltration, no denial of service. This is a defensive/AppSec tool. Unauthorized access is a crime in most jurisdictions (CFAA in the US, the Computer Misuse Act in the UK, equivalent computer-crime statutes elsewhere) and the operator carries that responsibility.
When to use it
Reviewing or pentesting an authorized codebase, service, binary, or cloud configuration — or triaging an existing scan down to what is actually exploitable.
Role mapping
| Core role | Security role |
|---|---|
| LEAD | Split by trust boundary / entry point / high-value asset; maintain the coverage map |
| BUILDER | HUNTER (clean context): propose candidates with an exploitation hypothesis (source → sink) |
| CRITIC | VALIDATOR (blind): inspect the real artifact, check reachability and taint, build the PoC. No PoC, no pass |
The bar (three axes, all required)
- Exploitability: a working PoC or equivalent evidence under the target's actual conditions — not "theoretically possible".
- CWE mapping (prefer the CWE Top 25): map to a meaningful weakness class, e.g. CWE-89, CWE-79, CWE-787, CWE-416, CWE-918, CWE-862, CWE-22, CWE-502.
- Real precedent (never fabricated): point to the real-world vulnerability class or a genuine CVE to calibrate severity and credibility. If you cannot find a precedent, write "novel/uncertain" — do not invent a CVE ID.
What ships with it
4 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.
- 11d ago First seen · 76 lines · 196 tokens per session scan A 26096f739f9f
security-vuln-gauntlet is a skill published in the GitHub repository trilwu/gauntlet-loop-skills (3 stars, last pushed 1mo ago), licensed MIT. It adds 196 tokens to every session and 1,301 once invoked, about $0.0010 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
prompt-injection-tester
Red-team an LLM application against prompt injection and jailbreaks using a curated, categorized payload library and canary-based detection, then produce a resilience score. Use when the user asks to "test my chatbot for prompt injection", "check if my AI app is jailbreakable", "red-team my LLM", "evaluate…
aidex-coverage
Use when writing, placing, or running tests in any project — which layer a behaviour belongs in ("unit or E2E for X", "component test or browser test"), which tests to run for a change instead of the whole suite, when to extract a fixture, setting up an isolated disposable E2E environment, or the per-project testing…
aidex-bugfix
Use when fixing a bug, resolving a reported issue, or when something is broken and needs a test-driven fix — investigate root cause, write a failing regression test (RED), implement the minimum fix, confirm the test passes (GREEN), then commit test and fix together. Fires on "fix this bug", "this is broken", "it's not…
dast-workflow
Dynamic Application Security Testing workflow — OWASP ZAP automation (baseline/full/API scans), Burp Suite Professional playbooks, Burp Collaborator for out-of-band detection, auth-state orchestration, and CI integration with scope-safe active scanning.
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
test-first-bugs
Enforces a test-driven bug-fixing workflow. Use when a user reports a bug, failing code, an error, or asks to fix something.