large-file-refactor
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted source code by scanning files in a user-provided directory. The agent is instructed to analyze symbol overviews, references, and history of these files. Malicious content within the scanned files could attempt to override the agent's instructions or bias the resulting refactor plan. \n
- Ingestion points: The helper script reads file contents to estimate lines of code, and the workflow directs the agent to manually inspect the candidate files. \n
- Boundary markers: There are no explicit delimiters or instructions to ignore embedded commands within the analyzed files. \n
- Capability inventory: The agent has access to file inspection tools and the ability to execute the refactor helper script. \n
- Sanitization: The skill does not filter or sanitize the content of the files before the agent analyzes them. \n- [COMMAND_EXECUTION]: The helper script invokes system CLI tools such as git, tokei, and rg via subprocess.run to perform its analysis. These calls use the list format to mitigate command injection risks but rely on the integrity of the external binaries found on the system.
Audit Metadata