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 oscarsterling/clelp-skills --skill gauntletgit clone --depth 1 https://github.com/oscarsterling/clelp-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/oscarsterling/clelp-skills/gauntlet)<a href="https://agentmods.dev/skills/oscarsterling/clelp-skills/gauntlet"><img src="https://agentmods.dev/badge/skills/oscarsterling/clelp-skills/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/oscarsterling/clelp-skills/gauntlet"><img src="https://agentmods.dev/badge/skills/oscarsterling/clelp-skills/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 | $0.00081 | $0.01759 |
| Opus 5 | $0.00041 | $0.00879 |
| Sonnet 5 | $0.00016 | $0.00352 |
| Haiku 4.5 | $0.00008 | $0.00176 |
Grade A, and why
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 10d 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 — 139 lines — stays where its author put it; the contents beside it link to each section on GitHub.
The Gauntlet
The Gauntlet is a loop for hardening code that must not fail silently: deletion guards, permission hooks, safety gates, anything where a single missed case is a real incident. It is more than "ask an AI to review this." Plain review gives you a list of vibes and maybe a real bug buried in it. The Gauntlet is structured so that the only things that survive are concrete, reproducible failures, and so that one model's blind spot does not become your shipped bug.
When to use it
Reach for the Gauntlet when all of these are true:
- A wrong result is expensive or irreversible (data loss, a security bypass, a destructive action taken on bad state).
- The failure surface is enumerable but large: many small cases, easy to miss one.
- You can write down a clear pass/fail contract for the code under review.
If the cost of being wrong is low, or you cannot articulate what "wrong" means, this loop is overkill. Use a normal review.
The five pillars
These are what separate the Gauntlet from ordinary red-teaming. Drop any one of them and the loop stops working.
-
The concrete-reproducer bar. A finding only counts if it comes with an input or a state that actually produces the failure: the exact commands, the exact file contents, the sequence of events. "This looks risky" is not a finding. "Here is a repo state where the gate passes and deletes a commit that exists nowhere else" is a finding. This bar is what filters style notes from real holes, and it is what makes a fix verifiable: you can reproduce the state, apply the fix, and watch the failure disappear.
-
Cross-lab two-model review. Send the same packet to two models from different labs (this skill ships an OpenAI leg and a Google Gemini leg). They have different blind spots. In a real run, one model confidently returned GO on the exact area where the other model proved a concrete hole. A single reviewer would have shipped that bug. The orchestrator adjudicates disagreements: when the models split, you read both arguments and decide which one reasoned correctly, rather than averaging them.
What ships with it
5 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.
- 10d ago First seen · 139 lines · 81 tokens per session scan A 807bff09ab82
gauntlet is a skill published in the GitHub repository oscarsterling/clelp-skills (0 stars, last pushed 4d ago), licensed MIT. It adds 81 tokens to every session and 1,759 once invoked, about $0.0004 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
research-engineer
An uncompromising Academic Research Engineer. Operates with absolute scientific rigor, objective criticism, and zero flair. Focuses on theoretical correctness, formal verification, and optimal implementation across any required technology.
tika-eval-compare
Compare extracts from two Tika builds over a corpus to detect regressions in content, encoding, exceptions, and embedded-document handling. Use for "compare before/after extracts", "eval this change against the corpus".
neuron-evaluation-engineer
Create and run AI evaluations with datasets, assertions, and output drivers in Neuron AI. Use this skill whenever the user mentions evaluation, testing AI systems, creating evaluators, dataset-driven testing, assertion-based validation, or wants to measure AI system performance. Also trigger for tasks involving…
jetson-validate-image
Use after jetson-flash-image to run static BSP checks, on-target smoke/regression tests on a flashed DUT, or both. Not for build or flash steps. Triggers: validate bsp, on-target validation.
atmos-validation
Validate Atmos projects, components, arbitrary JSON Schema inputs, EditorConfig, and GitHub Actions; use affected-file selection and native CI annotations.
skill-benchmark
Benchmark AI skill effectiveness by measuring implementation quality against legacy constraints.