The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides and executes several shell scripts (scan-all.sh, scan-npm.sh, scan-python.sh, scan-ci.sh) to audit the local development environment and project files. These scripts use standard utilities like grep, find, and pip to identify potential security risks.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from project files and reports it to the agent without sanitization.
Ingestion points: The scripts scan-npm.sh, scan-python.sh, and scan-ci.sh perform recursive searches and read content from various user-controlled files such as package.json, requirements.txt, and GitHub Actions workflows.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to ignore potentially malicious content embedded within the files being scanned.
Capability inventory: The skill allows the agent to execute complex shell-based scanning logic and interpret the results to provide remediation advice to the user.
Sanitization: No sanitization, escaping, or validation is performed on the content read from external files before the scan results are presented to the agent for interpretation.

supply-chain-guard