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 OKHP3/skillz --skill security-analysisgit clone --depth 1 https://github.com/OKHP3/skillzWrote 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/okhp3/skillz/security-analysis)<a href="https://agentmods.dev/skills/okhp3/skillz/security-analysis"><img src="https://agentmods.dev/badge/skills/okhp3/skillz/security-analysis/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/okhp3/skillz/security-analysis"><img src="https://agentmods.dev/badge/skills/okhp3/skillz/security-analysis.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 | $0.00098 | $0.01319 |
| Opus 5 | $0.00049 | $0.00660 |
| Sonnet 5 | $0.00020 | $0.00264 |
| Haiku 4.5 | $0.00010 | $0.00132 |
Grade A, and why
security-analysis 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 6d 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 — 113 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Security analysis
A finding is only real if you can describe the attacker, what they control, and what they get. "This could be unsafe" is not a finding — it is an unfinished one. The discipline here is identical to debugging: trace from a source you can influence to a sink that does something dangerous, and prove the path is unbroken.
Scope: this is for reviewing code you or the user are authorized to work on — your own projects, your employer's, or an engagement you have permission for. It finds and fixes defects. It does not produce working exploits against systems you do not control.
1. Map trust boundaries first
You cannot assess code without knowing where untrusted data enters. Identify:
- Sources: HTTP request bodies, query params, headers, cookies, uploaded files, webhook payloads, message queues, third-party API responses, filenames, environment in multi-tenant contexts, and anything read from the database that a user once wrote
- Sinks: SQL, shell, filesystem paths, HTTP requests made by the server, template
rendering, deserializers,
eval-alikes, redirects, log statements that feed a parser - The boundary: where a request stops being anonymous and becomes a principal with rights
That last one, written down explicitly, is where most real findings live.
Done when: you have listed the sources and sinks in the code under review.
2. Trace source to sink
For each sink, walk backwards. The question is never "is this validated?" but "is every path to here validated, including the one added last week?"
The path is broken, and there is no finding — only if a transformation makes the data safe for that specific sink. Safe for one sink is not safe for another: HTML-escaping does nothing for SQL, and SQL parameterization does nothing for a shell command.
Done when: each source→sink pair is either proven safe or recorded as a finding.
3. Check the categories that get missed
Injection is well known and usually handled. These are where real findings cluster:
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.
- 6d ago First seen · 113 lines · 98 tokens per session scan A 5b2256f2bc29
security-analysis is a skill published in the GitHub repository OKHP3/skillz (3 stars, last pushed yesterday), licensed MIT. It adds 98 tokens to every session and 1,319 once invoked, about $0.0005 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 2026-09-03.
Other skills, from other repositories
vastai-sdk
Vast.ai Python SDK — high-level API for GPU instances, volumes, serverless endpoints, and billing.
pm-aarrr
A post-launch product-growth workflow based on AARRR: acquiring users, activating them, retaining them, earning revenue, and gaining referrals.
pm-docs
A workflow for producing product documents such as PRDs, BRDs, and MRDs. These are structured documents describing what a product needs, its business case, and its market.
product-marketing-copywriter
A marketing-copy tool that analyzes audience problems and product benefits, then creates promotional headlines and body text. Marketing copy is writing intended to explain and promote a product.
pm-position
A guided process for defining a product’s market position, value, audience, competitive difference, business model, pricing, and revenue plan. The instructions are written mainly in Chinese.
pm-retro
A retrospective workflow for reviewing a finished agile iteration, recording what happened, and choosing improvements for the next cycle.