extract-source-sample
Pass
Audited by Gen Agent Trust Hub on Jun 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell commands to handle Git LFS assets and identify file types. Evidence: SKILL.md instructs the agent to run 'git lfs fetch', 'git lfs checkout', and 'file ' to process assets.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection by processing untrusted data from user-provided directories. Evidence: 1. Ingestion points: Data is read from 'working/script.json', 'HOW_TO_MAKE_THIS_VIDEO.md', and 'working/*.py' within the provided 'run-dir'. 2. Boundary markers: Content is read into the agent context without specific boundary markers or instructions to ignore potential commands embedded in the data. 3. Capability inventory: The agent can execute shell commands ('git lfs', 'file') and perform file system writes to the character library. 4. Sanitization: No sanitization is performed on the text read from ad-run files before it is used to populate the output JSON or drive character library updates.
- [EXTERNAL_DOWNLOADS]: The skill performs remote downloads using Git LFS. Evidence: SKILL.md directs the agent to fetch LFS pointers from remote servers if binary files are missing.
Audit Metadata