diffity-resolve
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses local CLI tools (
which diffity,npm install -g diffity, and variousdiffity agentsubcommands) to list, modify, and manage review comments. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input in the form of external review comments (
diffity agent list --status open --json) which could hypothetically contain instructions attempting to manipulate agent behavior. - Ingestion points:
SKILL.md(Step 1 under instructions reads JSON comment threads). - Boundary markers: Absent in instructions.
- Capability inventory: Invokes
diffityCLI subcommands and standard code editing capabilities via an Edit tool. - Sanitization: None explicitly mentioned in the static workflow definition.
Audit Metadata