blender-cinematography
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill demonstrates strong security posture by defining explicit capability boundaries. It specifically states that camera work does not authorize rendering, export, or downstream generation, preventing data exfiltration or unauthorized resource usage. Furthermore, it incorporates an acceptance checklist in
references/acceptance-checklist.mdthat requires authorization for any potential persistence or modification tasks. - [INDIRECT_PROMPT_INJECTION]: The skill processes external data such as subject lists and object names, which serves as an ingestion surface for potential indirect instructions. The risk is mitigated by strict workflow requirements that mandate reporting unverified inputs and using compact evidence packets for validation.
- Ingestion points: Subject lists and object definitions provided at runtime as described in
SKILL.md. - Boundary markers: Presence of structured request shapes and an acceptance checklist that requires explicit target and frame range definitions.
- Capability inventory: Limited to camera creation, lens configuration, and visibility validation tools; no network access or arbitrary command execution is permitted.
- Sanitization: Relies on strict adherence to input prerequisites and the rejection of invented or ambiguous data.
Audit Metadata