descript
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process and edit transcripts of audio recordings, which are external, untrusted inputs.
- Ingestion points: Processes transcripts of podcasts and interviews in
SKILL.mdandreferences/the-words-framework.md. - Boundary markers: Relies on manual human verification ('human verifies by ear') rather than technical input delimiters.
- Capability inventory: The agent can plan edits, trigger Underlord actions via API/MCP, and publish content via WoopSocial as described in
SKILL.md. - Sanitization: No explicit sanitization or filtering of transcript content is documented.
- [EXTERNAL_DOWNLOADS]: The skill connects to external services and APIs for media processing and distribution.
- Evidence: Mentions the Descript Public API, MCP connections for automated editing features, and publishing via WoopSocial in
SKILL.mdandreferences/descript-2026-reality.md.
Audit Metadata