ship-it
Pass
Audited by Gen Agent Trust Hub on May 19, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses the
$ARGUMENTSplaceholder inSKILL.mdto allow user-supplied content to be interpolated directly into the agent's instructions. This creates a vulnerability surface for indirect prompt injection where untrusted input could attempt to override the skill's logic or bypass safety constraints during execution. - Ingestion points: The surface is located in
SKILL.mdvia the$ARGUMENTSvariable. - Boundary markers: No delimiters, markers, or explicit instructions to ignore embedded commands are present around the interpolated content.
- Capability inventory: The skill provides the agent with the ability to modify the local git environment, including creating new branches, committing code changes, and opening pull requests.
- Sanitization: No validation, escaping, or filtering of the injected arguments is implemented within the skill logic.
Audit Metadata