supply-chain-security-scanner
Fail
Audited by Gen Agent Trust Hub on Jul 10, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill's documentation and internal scripts suggest installing audit tools by piping remote scripts from the official Anchore GitHub repository directly to the shell. Automated scans identified this as a remote code execution risk.
- [EXTERNAL_DOWNLOADS]: Retrieves vulnerability databases and provenance data from official remote repositories during operation.
- [COMMAND_EXECUTION]: Executes several security CLI tools locally (e.g., npm, pip-audit, syft, trivy) to perform file-level and image-level security audits.
- [DATA_EXFILTRATION]: Includes an optional feature to transmit SBOM reports to a user-configured OWASP Dependency-Track instance via curl.
- [PROMPT_INJECTION]: The skill processes untrusted metadata from project manifests and container images, creating an indirect prompt injection surface. Ingestion points: Files such as package.json, requirements.txt, and Dockerfiles. Boundary markers: No specific delimiters are used to prevent the agent from obeying instructions embedded in metadata. Capability inventory: Local shell execution, file system access, and network requests. Sanitization: Uses jq for parsing tool outputs to ensure structural integrity.
Recommendations
- HIGH: Downloads and executes remote code from: https://raw.githubusercontent.com/anchore/grype/main/install.sh, https://raw.githubusercontent.com/anchore/syft/main/install.sh - DO NOT USE without thorough review
Audit Metadata