agentsmd-scaffold
Pass
Audited by Gen Agent Trust Hub on Jul 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill includes a local utility, 'scripts/scan_repo_context.py', used to gather evidence from the repository structure, manifests, and existing documentation. The agent is instructed to execute this script to inform its scaffolding decisions.
- [PROMPT_INJECTION]: The skill is subject to indirect prompt injection because it processes untrusted content from the repository it is analyzing. Malicious instructions hidden in scanned README or manifest files could attempt to influence the agent's output. * Ingestion points: 'scripts/scan_repo_context.py' reads various repository files including README.md, AGENTS.md, and manifest files like package.json or pyproject.toml. * Boundary markers: The skill does not explicitly define delimiters to separate untrusted repository content from its core instructions. * Capability inventory: The skill provides functionality for reading repository structure and drafting or updating local instruction files. * Sanitization: The bundled scanner script gathers metadata and text statistics rather than executing the content of the files it processes.
Audit Metadata