# .claude/agents/security-reviewer.md
---
description: "Reviews code changes for security vulnerabilities"
allowed-tools: Read, Grep, Glob
model: opus
---
You are a security reviewer. Analyse the given code for:
- Injection vulnerabilities (SQL, NoSQL, command, XSS)
- Authentication/authorization flaws
- Sensitive data exposure
- Insecure dependencies

"Implement the user notifications feature:
1. Read the spec in docs/notifications-spec.md
2. Implement the backend endpoints
3. Write comprehensive tests
4. Review for security issues
5. Update the OpenAPI spec
6. Create the PR with a full description"

import { claude } from "@anthropic-ai/claude-code";

const result = await claude({
  prompt: "Run tests and fix any failures",
  workingDirectory: "./my-project",
  allowedTools: ["Bash", "Read", "Write", "Edit"],
  model: "sonnet"
});

# .github/workflows/claude-review.yml
- name: Claude Code Review
  uses: anthropics/claude-code-action@v1
  with:
    prompt: "Review this PR for bugs, security, and style"
    allowed_tools: "Read,Grep,Glob"

claude -p "Verify this branch is ready to merge:
  1. All tests pass
  2. No TODO/FIXME markers
  3. OpenAPI spec matches implementation
  4. No console.log statements in production code"
  --allowedTools Bash,Read,Grep

# Every morning at 9am — pull latest, run tests, report failures
claude -p "Pull main, run full test suite, and if anything fails, create a GitHub issue with the details"

# Weekly on Monday — dependency audit
claude -p "Run npm audit and create a PR fixing critical vulnerabilities"

# Nightly — code quality report
claude -p "Analyse code quality metrics and post a summary to #engineering Slack"