The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs users to install code from a third-party GitHub repository (sparkfinderoven/r01-hesreallyhim-awesome-claude-code-security) and references tools from ara.so, which are not verified vendor resources or established services.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data—such as source code, dependency files, and architecture descriptions—creating a vulnerability to indirect prompt injection.
Ingestion points: Files targeted by commands like /owasp-scan, /dep-cve, and /secret-detect.
Boundary markers: The skill documentation does not define delimiters or protective instructions for the ingested content.
Capability inventory: The suite has broad capabilities, including local command execution, file system writes, and network requests.
Sanitization: No validation or escaping of processed data is documented before it is used in reports or automated workflows.
[COMMAND_EXECUTION]: Provides a large set of commands that execute auditing and scanning tools directly on the user's host machine.
[DYNAMIC_EXECUTION]: Dynamically generates bash scripts for pre-commit hooks and includes automation scripts that use the subprocess module to run commands at runtime.
[DATA_EXFILTRATION]: Includes integrated functionality to send security findings and incident data to external platforms like Slack and Jira via webhooks and API calls.
[PERSISTENCE]: Sets up Git pre-commit hooks that persist in the repository to automatically run security checks during every commit.

sparkfinderoven-r01-security-compliance-skills