agentleFS
gener8vgener8v/gener8v.claude-skills/skills/owasp-top10-review/SKILL.md

owasp-top10-review

Skill3 starsChanged 2 months agoMITView on GitHub

Category-by-category OWASP Top 10:2025 coverage assessment of a codebase: maps existing SEC-XXX findings onto the taxonomy, probes the categories bottom-up reviews miss (supply chain, exceptional conditions, logging, SSRF) and re-ranks priorities. Use when asked 'are we OWASP-compliant' or 'how do we score against the Top 10', before an audit or a security questionnaire, or after security reviews accumulate. Not for one delivered ticket (security-review) or the model surface of an LLM application (owasp-llm-top10-review).

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 6 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill when:

    line 19
  2. Distinctive:

    After a Security Review, to organize findings against a recognized taxonomy and check for blind spots

    line 21
  3. Distinctive:

    Before an audit, certification, or customer security questionnaire

    line 22
  4. Distinctive:

    When prior security reviews or assessments exist and need re-ranking against the current OWASP edition (the 2025 list reshuffled and …

    line 23
  5. Distinctive:

    If per-ticket security reviews exist, map their SEC-XXX (qualified) onto categories rather than re-describing each finding.

    line 98
  6. Distinctive:

    Every Medium-or-higher finding states who the attacker is, what access they have, and what they achieve.

    line 104

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

OWASP Top 10 Review Skill

Purpose

Perform a systematic, category-by-category assessment of a codebase against the OWASP Top 10:2025 web application security risks. Where the Security Review skill hunts for individual vulnerabilities bottom-up, this skill works top-down: it walks all ten OWASP categories, assesses the system's posture in each, maps existing findings (SEC-XXX) onto the framework, verifies the categories a finding-by-finding pass tends to under-probe, and produces a coverage matrix that makes gaps and strengths legible against an industry-standard taxonomy.

It is the framework-coverage complement to Security Review: Security Review finds the bugs; this skill proves coverage and surfaces whole categories that were missed.

When to Use

Use this skill when:

  • A stakeholder asks "are we OWASP-compliant?" or "how do we score against the Top 10?"
  • After a Security Review, to organize findings against a recognized taxonomy and check for blind spots
  • Before an audit, certification, or customer security questionnaire
  • When prior security reviews or assessments exist and need re-ranking against the current OWASP edition (the 2025 list reshuffled and renamed several categories)
  • For periodic posture reviews where category completeness matters more than finding novelty

Input

Source: Delivered code, configuration, infrastructure-as-code, CI/CD, and any existing security review reports Read from:

  • The codebase (application routes, auth, data layer, config, containers, compose/IaC, CI workflows)
  • Every existing security review report — .gener8v/changes/*/reviews/*-security-review.md (one per delivered ticket, under the change it belongs to; a legacy project may still hold them at .gener8v/reviews/*-security-review.md) — to map their SEC-XXX findings onto categories rather than re-deriving them. Numbering restarts per report, so findings are referenced with the change segment, <change-slug>/<report-slug>/SEC-XXX — e.g. support-search/search-and-retrieval-ticket-002-security-review/SEC-002 (CONVENTIONS.md §4)
  • Constraints: .gener8v/constraints/[slug].md (if available — for compliance requirements)
  • Technical Design / System Context (if available — for deployment topology)

Expects: Code to exist. Degrades gracefully — with no prior security review it derives findings directly; with one, it maps and supplements.

Output

Produces: An OWASP Top 10:2025 coverage assessment Write to: .gener8v/reviews/[slug]-owasp-top10-assessment.md Creates directory: .gener8v/reviews/ if absent Naming: -owasp-top10-assessment suffix

New findings discovered during the pass are numbered SEC-001… within this assessment and described in full here; other documents reference them as <system-slug>-owasp-top10-assessment/SEC-XXX. Per-ticket security reviews are point-in-time reports and are never edited by this skill. There is no shared register to maintain: Orchestrate lists every review and assessment, and the qualified ID is what makes a reference unambiguous.

The Framework — OWASP Top 10:2025

Assess every category. The 2025 edition (finalized January 2026) reshuffled and added categories — use these, not the 2021 list:

  • A01:2025 — Broken Access Control (authn, authz, IDOR, tenant isolation, function/object-level checks)
  • A02:2025 — Security Misconfiguration (↑ from #5: env gating, exposed surfaces, CORS, headers, container hardening, IaC, default secrets)
  • A03:2025 — Software Supply Chain Failures (NEW, broadens "Vulnerable & Outdated Components": dependencies + build systems + distribution + CI/CD integrity)
  • A04:2025 — Cryptographic Failures (hashing, token signing, data-at-rest/in-transit, key management)
  • A05:2025 — Injection (SQL/NoSQL/command/template/LDAP, XSS, insecure deserialization)
  • A06:2025 — Insecure Design (design-level flaws, missing controls by design, abuse cases, mass assignment, resource limits)
  • A07:2025 — Authentication Failures (credential strength, brute-force protection, session/token lifecycle, enumeration)
  • A08:2025 — Data Integrity Failures (deserialization integrity, update integrity, CI/CD pipeline integrity not covered by A03)
  • A09:2025 — Security Logging & Alerting Failures (logging of auth/authz events, monitoring coverage, alerting)
  • A10:2025 — Mishandling of Exceptional Conditions (NEW, 24 CWEs: fail-open behavior, improper error handling, failing into insecure state)

Note: SSRF is no longer a standalone top-level category in 2025 (folded in) — still assess it as a sub-check under A10/A05.

Output Format

# [System]: OWASP Top 10:2025 Assessment

## 1. Coverage Summary
[Table: # | Category | Status (Adequate / Gaps / Primary gap / Not applicable) | Driving findings]
[2-3 sentence net: where risk concentrates, which categories pass.]

## 2. Category-by-Category Assessment
### A01 — Broken Access Control · [status]
**What was checked:** ...
**Evidence (controls present):** ...
**Gaps / findings:** [mapped SEC-XXX + any new]
[... repeat for A02–A10 ...]

## 3. New-Category / New-Finding Pass
[Focused analysis of A03 supply chain and A10 exceptional-conditions, plus any new SEC-XXX raised, fully described: severity, category, location, attack scenario (Med+), impact, recommendation.]

## 4. Finding Totals
[Counts by severity; table of findings by 2025 category slot.]

## 5. Conclusion
[Priority guidance re-ranked by 2025 weighting.]

Principles

Assess Every Category — Including "Not Applicable"

A category with no findings is a result, not a skip. State what was checked and why it does not apply (e.g. "no vector DB → A10-LLM N/A"). A blank category reads as "not assessed," which is the opposite of coverage.

Verify the Under-Probed Categories

Bottom-up bug hunts reliably miss A08 (insecure deserialization), A09 (logging gaps), A10 (fail-open error handling), and the SSRF sub-check. Actively grep for them: pickle/yaml.load/marshal (deserialization); logging calls around auth failures and authz denials (A09); broad/except Exception and silent fallbacks (A10); backend fetches of user-controlled URLs (SSRF). Absence of evidence is only evidence of absence after you look.

Map, Don't Duplicate

If per-ticket security reviews exist, map their SEC-XXX (qualified) onto categories rather than re-describing each finding. Add value through coverage and re-ranking, not restatement. Only fully describe findings this pass newly discovers.

Honor the 2025 Re-Ranking

The 2025 edition moved Security Misconfiguration to #2 and introduced Software Supply Chain Failures at #3. When advising priority, reflect the current weighting — a misconfiguration cluster is now a #2-class concern, not a #5-class one.

Attack Scenarios for Medium+

Every Medium-or-higher finding states who the attacker is, what access they have, and what they achieve. A finding without a scenario is an assertion.

Compliance Constraints Are Non-Negotiable

A documented compliance constraint (CC-XXX) violation is automatically Critical regardless of exploitability.

Process

  1. Locate inputs — read the codebase and every existing security review under .gener8v/changes/*/reviews/ (and the legacy .gener8v/reviews/*-security-review.md, if present).
  2. Inventory the attack surface — routes/entry points, auth model, data layer, config, containers, IaC, CI/CD.
  3. Walk A01→A10 — for each, record what was checked, evidence of controls present, and gaps; map existing SEC-XXX.
  4. Run the under-probed verifications — deserialization, security-event logging, fail-open except patterns, SSRF, mass assignment. Use targeted greps; cite file:line.
  5. Deep-pass the 2025 additions — A03 (deps + base-image pinning + CI integrity + a continuous-scan gate) and A10 (enumerate every degrading/broad except; classify fail-open vs fail-closed vs intentional-degradation; the security-relevant fail-opens become findings).
  6. Raise & number new findings — SEC-001… within this assessment; full descriptions; findings that need code changes become tickets via Ticket Breakdown (cite the qualified ID).
  7. Build the coverage matrix + re-ranked priority — counts by severity, findings by 2025 category.
  8. Write the report to .gener8v/reviews/.

Example (abbreviated)

A02 — Security Misconfiguration · Primary Gap. Placeholder database password in a fallback connection string (support-search/search-and-retrieval-ticket-002-security-review/SEC-001), root containers and unpinned base images (new findings SEC-001, SEC-002 in this assessment), CORS wildcard methods/headers and missing security headers (SEC-003, SEC-004). Elevated to #2 under the 2025 weighting.

A10 — Mishandling of Exceptional Conditions · Minor. Reviewed every degrading except: a missing DATABASE_URL now fails closed (correct, after the ticket-002 fix); the ingestion worker falls back to shared source credentials on a decrypt error (fail-open → new finding SEC-005); indexing-queue degradations are intentional but should alert (ties A09 / …ticket-002-security-review/SEC-002).

Troubleshooting

  • The assessment covered the wrong system, or none of the focus the user asked for. This skill runs in a forked subagent that sees the skill and its arguments, not the conversation. Put the system slug, and any surface to weight (the public API, the admin console), in the argument. The slug names the report .gener8v/reviews/<slug>-owasp-top10-assessment.md; Orchestrate finds assessments in .gener8v/reviews/ by their -assessment.md suffix, so keep the name exact.
  • Two different findings are both SEC-001. Numbering restarts per report, and this assessment starts its own. Always cite qualified IDs — support-search/search-and-retrieval-ticket-002-security-review/SEC-002 for a ticket review, <slug>-owasp-top10-assessment/SEC-001 for this one — in the matrix, the priorities and any ticket that fixes a finding.
  • The A03 deep pass cannot run a dependency or image scan. Say which scan was not run and why, and rate A03 on what was verified by reading — pinning, lockfiles, CI integrity. Do not mark it clean on the strength of a scan that never happened.
  • The system also calls a model. Prompt injection, output handling and denial-of-wallet belong to the OWASP LLM Top 10 Review, run alongside. Do not fold them into A03 or A05 here; note that the model surface is assessed there.

Integration with Other Skills

Upstream: Security Review (provides the per-ticket SEC-XXX findings to map, under changes/<change-slug>/reviews/); Constraints (compliance); Technical Design (topology). Parallel: OWASP LLM Top 10 Review (for AI/LLM systems — run both; this covers the web surface, that covers the model surface). Downstream: Audit (consumes coverage results); Ticket Breakdown (new findings that need code changes become tickets); Orchestrate (lists the assessment under cross_cutting.assessments).

Revisions

  • Re-run when a new OWASP edition lands (re-rank the findings), after deliveries that touch authentication, configuration, dependencies or external input, and before an audit or questionnaire.
  • Re-running replaces [slug]-owasp-top10-assessment.md; carry forward the previous assessment's open findings with their status so nothing is silently dropped.

Notes

  • Point-in-time. OWASP revises the Top 10 periodically (2021 → 2025); confirm you are assessing against the current edition and re-rank prior assessments when a new edition lands.
  • This skill assesses against a taxonomy — it does not replace deep bug-hunting (Security Review) or threat modeling.
  • For API-heavy systems, add a short OWASP API Security Top 10 (2023) cross-map — it resolves object/function-level authz and resource-consumption risks the web list blurs.
  • Can run standalone (derives findings) or as a coverage layer over an existing Security Review (maps + supplements).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.