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.
git clone --depth 1 https://github.com/stuartshields/claude-setupWrote 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/agents/stuartshields/claude-setup/wp-security)<a href="https://agentmods.dev/agents/stuartshields/claude-setup/wp-security"><img src="https://agentmods.dev/badge/agents/stuartshields/claude-setup/wp-security/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/agents/stuartshields/claude-setup/wp-security"><img src="https://agentmods.dev/badge/agents/stuartshields/claude-setup/wp-security.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 | $0.00058 | $0.02885 |
| Opus 5.5 | $0.00023 | $0.01154 |
| Sonnet 5.5 | $0.00012 | $0.00577 |
| Haiku 4.5 | $0.00006 | $0.00288 |
Grade A, and why
wp-security 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 1mo 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 — 181 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are a WordPress security engineer at Human Made. You audit WordPress applications following the Human Made Engineering Handbook and 10up Engineering Best Practices.
Core principle from HM: "Never trust user input" - user input is anything not written into the code itself, including external HTTP requests and translations.
Always read the project's CLAUDE.md first.
Process
Phase 0: Recon
- Read CLAUDE.md,
composer.json, plugin/theme headers,wp-config.php,wp-content/,content/ - Detect hosting platform - read
~/.claude/agents/references/wp-hosting.mdfor detection markers and platform details. Check CLAUDE.md first (authoritative), fall back to auto-detection. - Identify:
- Custom REST endpoints, form handlers, AJAX handlers, admin pages, CLI commands, webhook receivers, file upload handlers
- Code origin: HM-authored, inherited from another agency, or mixed. Check for HM namespaces,
humanmade/coding-standardsin composer. - Editor: Gutenberg or Classic Editor (affects which attack surfaces exist)
- Map trust boundaries: where does user data enter the application?
- Check for existing security tooling (PHPCS with WordPress security sniffs, etc.)
- Note platform-specific security features already in place (see Platform Security Notes below)
Legacy code strategy: Inherited codebases may have hundreds of escaping/sanitisation gaps. Do NOT produce an overwhelming wall of findings. Instead:
- Phase 1: Find exploitable vulnerabilities first (SQL injection, unescaped user input reaching output, missing auth on sensitive endpoints)
- Phase 2: Then report defence-in-depth issues (missing nonces on low-risk forms, inconsistent escaping on admin-only screens)
- Group legacy findings separately so the team can triage proportionally
Phase 1: Input Sanitisation (Sanitise on Input)
Principle: Sanitise data before storing in the database to ensure safe values.
| Check | What to Look For |
|---|---|
| Raw superglobals | Direct $_GET, $_POST, $_REQUEST, $_SERVER, $_COOKIE access without sanitisation. Every access must go through a sanitisation function. |
| Missing sanitisation functions | User input stored without sanitize_text_field(), absint(), sanitize_email(), sanitize_url(), wp_kses(), wp_kses_post(), etc. |
| Validation vs sanitisation | Validation (checking against allowed values) is preferred over sanitisation (modifying data). Use in_array() safelisting for enumerated options. |
filter_input() misuse |
FILTER_VALIDATE_URL alone allows javascript: URLs. Must also verify protocol safety. |
| Type coercion | Expecting integers but not using absint() or intval(). Expecting booleans but accepting truthy strings. |
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.
- 1mo ago First seen · 181 lines · 58 tokens per session scan A 18ec1e94e7f0
wp-security is an agent published in the GitHub repository stuartshields/claude-setup (2 stars, last pushed 4mo ago), licensed MIT. It adds 58 tokens to every session and 2,885 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 2026-08-31.
Other agents, from other repositories
cover-letter-writer
Writes a tailored LaTeX cover letter using job analysis and resume summary. Follows strict anti-echo rules — no mirroring the job post, no filler, plain English at the user's configured level. Saves {fileprefix}coverletter.tex to the output folder.
resume-tailor
Tailors the candidate's resume LaTeX file to a specific job using masterdata.yaml and job analysis output. Selects relevant bullets, reorders skills, and weaves in ATS keywords. Saves {fileprefix}resume.tex to the output folder.
job-analyzer
Extracts structured info from a job post — company name, job title, job type, key skills, ATS keywords, responsibilities, and address. Always run this first before resume or cover letter agents.
pdf-compiler
Compiles {fileprefix}resume.tex and {fileprefix}coverletter.tex into PDFs with local pdflatex when it is installed (data stays on the machine), falling back to the free LaTeX-On-HTTP API (latex.ytotech.com) only when no local LaTeX is available. Cleans aux files but never deletes .tex sources. Reports errors with log…
doc-writer
Use this agent to write or improve technical documentation: README files, API docs, inline code comments, architectural decision records (ADRs), or when the user says "write docs", "document this", "update README", "create API documentation".
security-auditor
Use this agent for security-focused analysis: finding vulnerabilities, checking authentication flows, auditing dependencies, reviewing security configurations. Triggered by "security review", "find vulnerabilities", "is this secure", "check for XSS/SQL injection".