difit-review
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exposes an indirect prompt injection surface by processing untrusted data from diffs and external repositories.
- Ingestion Points: Ingests local git revisions, GitHub URLs, or patch files specified by the user within SKILL.md.
- Boundary Markers: Absent; no explicit delimiters or instructions are used to segregate the diff content from the prompt structure.
- Capability Inventory: Executes shell commands via the
difitornpx difitcommands within SKILL.md. - Sanitization: Absent; the skill does not define methods to filter or escape code content before handling it.
Audit Metadata