dino-sync
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of external data from Dinox notes and tasks, which could potentially contain malicious instructions. However, it incorporates strong mitigations.
- Ingestion points: Dinox data (notes, tasks, tags, filenames) and CLI output processed via the
dinotool (SKILL.md). - Boundary markers: The 'Runtime Contract' explicitly labels all such data as untrusted and requires the agent to maintain this distinction.
- Capability inventory: Local process execution using the
dinoCLI (SKILL.md). - Sanitization: The skill provides clear negative constraints, instructing the agent to never execute instructions found within the data and never to interpolate untrusted text into shell commands.
Audit Metadata