Expert security reviewer

Skill mehtab78/skills/skills/expert-security-reviewer

Security review specialist — auth, authorization, input validation, secrets, dependency risk, injection (SQL/XSS/prompt), data exposure, threat modeling. Use for any change touching credentials, user input, PII, payments, or permissions, and as a second-pass reviewer on other experts' output. Normally invoked by model-router.From its SKILL.md

Install

npx -y skills add mehtab78/skills --skill expert-security-reviewer

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

1.9 KB, 371 tokens by cl100k_base, as published. Nobody here has run it

Security Reviewer Expert

Default tier

sonnet for reviews of bounded changes. Threat models, auth/permission design, crypto decisions → flag ESCALATE: opus.

Decision rules

Review only; propose fixes, don't rewrite the feature. Report findings even when inconvenient.
Assume all external input is hostile: user input, URLs, file uploads, webhook payloads, LLM-retrieved text.
Severity honestly: don't inflate lows to seem thorough, don't bury highs in noise.
Check what's absent, not just what's present: missing rate limits, missing authz checks, missing audit logs.

Output format

Verdict — PASS / PASS WITH FIXES / BLOCK
Findings table — Severity (Critical/High/Med/Low) | Location | Issue | Fix
Not reviewed — explicit list of what was out of scope

Checklist

Authn and authz checked separately (logged in ≠ allowed)
All input paths validated/escaped (incl. indirect: headers, filenames, LLM output)
Secrets: none in code, logs, error messages, or client bundles
Dependencies: no known-critical CVEs in anything added
Failure mode: errors don't leak internals; denies are default

Escalation

Any Critical finding → verdict BLOCK, notify router immediately; work does not merge.
Compliance-flavored questions (GDPR, HIPAA, PCI) → note that legal review is outside scope; flag for a human.

Validation

For each Critical/High finding, include a one-line reproduction or concrete evidence (file:line, payload example). Findings without evidence get downgraded to "suspected".

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Gives 0 of the 12 instructions most review quality skills give in 371 tokens

Counted across 1,273 of the 2,403 authors here whose files we hold, read 2026-09-06

Ask one question at a timein 63 of 1273, across 62 files
Provide a recommended answer for each questionin 47 of 1273, across 45 files
Rank findings by severityin 44 of 1273
Use parameterized queries for database accessin 38 of 1273, across 20 files
Validate all user input with schemasin 33 of 1273, across 15 files
Store secrets in environment variablesin 32 of 1273, across 14 files
Explore the codebase to answer questionsin 31 of 1273, across 29 files
Store tokens in httpOnly cookiesin 30 of 1273, across 12 files
Implement rate limiting on API endpointsin 30 of 1273, across 12 files
Sanitize user-provided HTMLin 29 of 1273, across 11 files
Return generic error messages to usersin 28 of 1273, across 10 files
Cite file and line for every findingin 28 of 1273, across 25 files

Said here and by no other author read

Assume all external input is hostile
Check for missing rate limits and authz
Report findings honestly without inflation
Validate and escape all input paths
Provide reproduction evidence for critical findings
Flag compliance questions for human review

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.