The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Recommends the installation and use of industry-standard security tools from reputable sources, such as semgrep, bandit, gitleaks, trufflehog, checkov, and tfsec. It also suggests utilizing verified GitHub Actions from organizations like aquasecurity and returntocorp for CI/CD security scanning.
[COMMAND_EXECUTION]: The skill uses the Bash tool to perform security reconnaissance (e.g., nmap, dig), execute SAST tools, and audit infrastructure configurations (e.g., aws, az, gcloud, kubectl). These operations are consistent with the skill's stated purpose as a security reviewer.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted third-party code and configuration files. While this constitutes an attack surface for indirect prompt injection, the skill mitigates this by mandating manual human review and providing specific guidance on identifying malicious injection patterns within the analyzed data.
Ingestion points: Local filesystem access via Read, Grep, and Glob tools to process project source code as described in SKILL.md.
Boundary markers: The skill does not define explicit delimiters but emphasizes manual verification and strict output reporting structures.
Capability inventory: Access to Bash for tool execution and various filesystem tools (Read, Grep, Glob) for data retrieval across all reference files.
Sanitization: Explicitly instructs the agent to recommend secure coding practices, such as using execFile and spawn to prevent command injection in the codebases it audits.
[CREDENTIALS_UNSAFE]: Contains example patterns and grep commands for detecting hardcoded secrets (e.g., AWS Access Keys, JWT tokens) in references/secret-scanning.md. These are documented as templates for scanning purposes and do not contain actual sensitive credentials.

security-reviewer