The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts scripts/agent.py and scripts/process.py execute external security tools (Checkov, tfsec) using the subprocess.run function. These calls use argument lists rather than shell strings, which is a secure implementation that prevents command injection vulnerabilities from untrusted file paths or directory names.
[EXTERNAL_DOWNLOADS]: The skill references and provides instructions for installing well-known DevSecOps tools from trusted registries (PyPI) and official GitHub Actions from established security vendors such as Bridgecrew and Aqua Security. These are standard, legitimate resources for infrastructure security.
[METADATA_POISONING]: There is a minor inconsistency between the author name 'mahipal' listed in the skill metadata and the 'mukul975' name found in the license and platform context. This discrepancy appears to be a common artifact of template usage rather than a deceptive attempt to misrepresent the skill's origin or functionality.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and analyze local infrastructure code files. This activity is performed using local binaries with constrained execution parameters. Evidence Chain: 1. Ingestion points: The scripts read files with .tf extensions and directories specified via CLI arguments. 2. Boundary markers: Not applicable as the skill functions as a wrapper for local security binaries. 3. Capability inventory: File system read/write for report generation and subprocess execution for scanning. 4. Sanitization: All system interactions utilize sanitized paths and list-based command construction, preventing data from being interpreted as executable instructions.

implementing-infrastructure-as-code-security-scanning