Skip to content

JPeetz/agent-skills / skills/security/security-and-hardening

security-and-hardening

Verified

Use this skill when auditing code for vulnerabilities, handling user input, authentication, data storage, or external integrations. Hardens code against OWASP Top 10 with a three-tier boundary system, dependency auditing, data privacy, and LLM safety.

FileKindSize
CHANGELOG.mdmarkdown501 B
LICENSEother665 B
SKILL.mdskill md9.0 KB
evals/evals.jsonconfig2.3 KB

evals/evals.json

View on GitHub
[
  {
    "id": "ev001",
    "description": "SQL injection prevention — agent identifies and fixes string concatenation in queries",
    "scenario": "A codebase has `SELECT * FROM users WHERE email = '` + userInput + `'`. Agent must identify the vulnerability and fix it with parameterized queries.",
    "expected_outcome": "Agent flags the SQL injection vector, replaces with parameterized query (e.g., `WHERE email = ?`), and validates the fix"
  },
  {
    "id": "ev002",
    "description": "Secrets management — agent identifies hardcoded secrets in source code",
    "scenario": "A commit adds a config file with `API_KEY = 'sk-abc123...'`. Agent must flag this as a security issue, recommend environment variable usage, and advise rotating the leaked key.",
    "expected_outcome": "Agent identifies the hardcoded API key, recommends env var injection, advises immediate rotation since the key is in git history"
  },
  {
    "id": "ev003",
    "description": "SSRF protection — agent identifies and mitigates user-supplied URL fetching",
    "scenario": "A webhook handler takes a URL from the request body and fetches it with `requests.get(url)`. Agent must identify the SSRF vector and implement an allowlist-based fix.",
    "expected_outcome": "Agent flags the SSRF vulnerability, implements URL allowlist with scheme and host validation, rejects private/reserved IPs, and forbids redirects"
  },
  {
    "id": "ev004",
    "description": "LLM output safety — agent recognizes model output as untrusted input",
    "scenario": "An app passes LLM output directly into `innerHTML` to render a chat response. Agent must identify the XSS vector and fix it with proper encoding.",
    "expected_outcome": "Agent flags that LLM output is untrusted, replaces innerHTML with framework-safe escaping or sanitization, and notes prompt injection risk"
  },
  {
    "id": "ev005",
    "description": "Data classification and retention — agent identifies missing privacy controls",
    "scenario": "A signup form collects name, email, and date of birth. There's no privacy notice, no TTL on the data, and no delete endpoint. Agent must identify the gaps.",
    "expected_outcome": "Agent flags missing data classification, missing retention limit, missing deletion endpoint, and recommends purpose specification and consent collection"
  }
]
rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first.Stars: 194kUpdated Claude CodeCodexVerifiedInstall