jt-review-work
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local git and gh commands via a bundled Python script (scripts/landed.py). The script uses safe subprocess invocation with list-based arguments and no shell. It also validates the task ID input against a strict numeric/dot pattern to prevent injection.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to read and display content from external sources (PRs, tasks, plan files). It explicitly instructs the agent to treat this content as data rather than instructions, mitigating the risk of instructions embedded in the project files influencing agent behavior.\n- [SAFE]: The skill is configured to disable autonomous model invocation and specifies that it is intended for human-only interaction. Its design adheres to the principle of least privilege and includes clear warnings for the agent regarding untrusted input.
Audit Metadata