poly-adk-conversations
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is a documentation resource for the PolyAI ADK suite. All external resources and commands mentioned are part of the official vendor infrastructure and intended for platform administration.
- [EXTERNAL_DOWNLOADS]: The skill references the installation of the 'polyai-adk' package. This is the official tool provided by PolyAI.
- [COMMAND_EXECUTION]: The instructions detail the use of the 'poly' CLI tool to list conversations and download audio recordings. These are legitimate development and debugging operations for the PolyAI platform.
- [INDIRECT_PROMPT_INJECTION]: The skill describes processes that involve reading conversation transcripts, which represents an ingestion surface for untrusted external data.
- Ingestion points: Conversation transcripts are accessed via the 'poly conversations get' command in SKILL.md.
- Boundary markers: No explicit boundary markers or instructions to ignore embedded content are specified.
- Capability inventory: The skill facilitates metadata retrieval and file downloads using the 'poly' CLI tool.
- Sanitization: There are no documented sanitization or filtering steps for the ingested transcript content.
Audit Metadata