Borrowing it
Nothing to install: this file belongs to adrienm7/ergopti. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/adrienm7/ergopti/main/.agents/skills/adversarial-audit/SKILL.mdgit clone --depth 1 https://github.com/adrienm7/ergoptiWrote 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/adrienm7/ergopti/adversarial-audit)<a href="https://agentmods.dev/skills/adrienm7/ergopti/adversarial-audit"><img src="https://agentmods.dev/badge/skills/adrienm7/ergopti/adversarial-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/adrienm7/ergopti/adversarial-audit"><img src="https://agentmods.dev/badge/skills/adrienm7/ergopti/adversarial-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 | $0.00062 | $0.01537 |
| Opus 5.5 | $0.00025 | $0.00615 |
| Sonnet 5.5 | $0.00012 | $0.00307 |
| Haiku 4.5 | $0.00006 | $0.00154 |
Grade A, and why
adversarial-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 15d 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 — 118 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Adversarial audit
Adopt an adversarial stance: the goal is not to confirm the code works, it is to find the input, state or timing that breaks it.
Read only the matching scope reference, plus the report contract before writing the deliverable:
- AutoHotkey: references/ahk.md
- Hammerspoon: references/hammerspoon.md
- Output: references/report-contract.md
This skill audits and reports. Do not edit production or test code unless the user separately asks to implement the findings.
The five guarantees
Audit against these, and state which one each finding violates:
- G1 — Robustness. No unhandled exception, crash or keyboard lock-up in any state: normal, suspended, paused, mid-layout-switch, mid-config-reload, updater-poll, cold start, first boot.
- G2 — No missing output. No silent no-op: every user action produces its effect or a loud failure.
- G3 — No races. Every async boundary (timer, hook, InputHook, HTTP, watcher) is safe against reentrancy and out-of-order completion.
- G4 — No lag. No perceptible latency on any hot path.
- G5 — Truthful preview. A tooltip or preview predicts the exact behavior the owning engine would produce in the same state.
Method
- Map the entry points — every hotkey, hook, timer, watcher, menu action, external trigger.
- Audit module by module, then flow by flow end-to-end. Module-local review misses the interaction bugs, which are the expensive ones.
- Question every async boundary. What happens if it fires twice? Out of order? While suspended? Before init? After teardown?
- Loop until dry. Keep sweeping until two consecutive passes surface nothing new. A fixed finding count means you stopped early, not that the code is clean.
Bug classes worth a dedicated sweep: unguarded map/table access, state read before init, load-time vs runtime registration, missing suspend guards, swallowed exceptions, non-reentrant handlers, sync I/O on a hot path. The scope references add platform-specific failure modes and current refutations.
What ships with it
3 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.
- 15d ago First seen · 118 lines · 62 tokens per session scan A d931ccebd876
adversarial-audit is a skill published in the GitHub repository adrienm7/ergopti (14 stars, last pushed yesterday), licensed MIT. It adds 62 tokens to every session and 1,537 once invoked, about $0.0002 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 2026-10-14.
Other skills, from other repositories
systematic-debugging
4-phase root cause debugging: understand bugs before fixing.
observability-langsmith
LLM observability platform for tracing, evaluation, and monitoring. Use when debugging LLM applications, evaluating model outputs against datasets, monitoring production systems, or building systematic testing pipelines for AI applications.
reenable-test
Guide for investigating, fixing, verifying via TDD, and re-enabling disabled or flaky Chromium tests (C++, Android Java, or WebUI). Use this skill when asked to deflake, fix, or re-enable a disabled or flaky test case, or to verify post-submit stability of a re-enabled test. Don't use for disabling a failing test…
experimental-code-coverage-local-debugger
Runs code coverage locally via Universal Test Runner (UTR) or helper scripts, mimicking LUCI trybots. Activate when CQ tryjobs fail or underreport coverage, to test local GN/recipe repairs before uploading, or to debug hermetic crashes.
verify-omnigent
Drive Omnigent the way a user does and prove a behavior with recorded evidence, using an isolated server, runner, and mock model plus a feature map of every user entry point. Load before reproducing a user-facing bug, before claiming a fix works, or when reviewing whether a change covered every surface (session vs.…
cli-e2e
Write, modify, or debug Docker-based Composio CLI end-to-end tests under ts/e2e-tests/cli, including binary invocation, fixture isolation, output assertions, and package manifests. Use for CLI E2E test suites only; use cli-command for CLI source implementation.