ワンクリックで
security-audit
Scan code for security vulnerabilities (OWASP patterns, path traversal, injection risks)
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
メニュー
Scan code for security vulnerabilities (OWASP patterns, path traversal, injection risks)
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Check and fix Checkstyle and PMD violations
Implement a GitHub issue with branch and pull request workflow
Post a status comment on a GitHub issue (starting work, plan update, completed)
Check JaCoCo code coverage
Format, validate checkstyle/PMD, and run tests before committing
Create/update GitHub issue, branch, commit, PR, then continue implementing
| name | security-audit |
| description | Scan code for security vulnerabilities (OWASP patterns, path traversal, injection risks) |
| argument-hint | ["file path"] |
| allowed-tools | Bash(./mvnw *), Read, Glob, Grep |
Scan the codebase (or specific file) for common security vulnerabilities.
If $ARGUMENTS is provided, limit the scan to that path. Otherwise, scan all source code.
Search for file path construction from untrusted input:
Grep for: new File(.*getName|Path.of(.*input|Paths.get(.*input
Verify each hit has canonical path validation or sanitization.
Search for Runtime.exec, ProcessBuilder with user input:
Grep for: Runtime.getRuntime|ProcessBuilder|\.exec\(
Search for unsafe YAML parsing:
Grep for: new Yaml\(\)|Yaml.load\(|ObjectMapper.*readValue.*untrusted
Verify Jackson is configured safely.
Search for XML parsing without disabling external entities:
Grep for: DocumentBuilder|SAXParser|XMLReader|TransformerFactory
Search for credentials, tokens, or keys in code:
Grep for: password|secret|token|apiKey|private.key (case-insensitive, exclude test files)
Check the SSL ignore feature (--ignoreSslErrors) for proper scoping:
| Severity | Location | Issue | Recommendation |
|---|---|---|---|
| CRITICAL | file:line | Description | Fix suggestion |
Report only confirmed findings, not theoretical risks. If clean, say "No vulnerabilities found."