The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze entire codebases, which are untrusted external data sources. This creates an attack surface where malicious instructions could be embedded within the audited code to attempt to override the agent's logic or bias its security reports.
Ingestion points: Source code files, configuration files, and dependency manifests within the target project directory (SKILL.md).
Boundary markers: The instructions do not specify any boundary markers or "ignore" directives to help the agent distinguish between its audit instructions and the content of the files it is reading.
Capability inventory: The skill utilizes shell-based tools including semgrep, npm audit, pip-audit, govulncheck, cargo audit, and trivy (SKILL.md).
Sanitization: There is no explicit sanitization or filtering of the code content described before the agent performs its manual or automated review.
[COMMAND_EXECUTION]: The skill provides instructions for the agent to execute several command-line security tools to automate the audit process. These include semgrep --config auto --json for static analysis and various ecosystem-specific auditors like npm audit, pip-audit, and trivy. These commands are standard and appropriate for the skill's primary stated purpose of security auditing.

vulnerability-scan