jt-drop-card
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes content from
card-loop/board.mdto identify task IDs and reasons for dropping them. While this constitutes a data ingestion surface, the risk is minimal due to the specific format requirements. - Ingestion point:
card-loop/board.md(reading existing board lines) - Boundary markers: None explicitly defined to separate data from instructions
- Capability inventory: Git branch checkout, Git commit operations, and file writing to
card-loop/board.md - Sanitization: No explicit sanitization of the 'reason' string appended to the file
- [COMMAND_EXECUTION]: The skill describes a workflow involving Git commands (checkout and commit). These operations are restricted to the local repository and specific files, aligning with the skill's purpose for project management automation.
- [SAFE]: The use of Thai language instructions for identifying tasks and modifying status lines is functional and does not contain any detected obfuscation or malicious intent. The skill includes safety checks, such as stopping if the working tree is dirty.
Audit Metadata