assemblyai-debug-bundle
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill mandates robust data protection measures by requiring the redaction of API keys, audio files, and transcripts from diagnostic evidence bundles before they are shared.
- [SAFE]: Authentication guidelines correctly identify best practices, such as using approved secret managers for API keys and strictly prohibiting the exposure of credentials in logs, URLs, or output.
- [SAFE]: External references target official documentation for AssemblyAI, a well-known technology provider, which is considered a safe source.
- [INDIRECT_PROMPT_INJECTION]: The skill processes repository code and log data, representing a potential surface for indirect prompt injection, which is addressed through specific redaction workflows.
- Ingestion points: Repository code, configuration, fixtures, and log files are inspected by the agent (SKILL.md).
- Boundary markers: The instructions include requirements to label content-bearing fields and replace sensitive data with fixed markers to separate data from instructions.
- Capability inventory: The skill utilizes file system tools (Read, Glob, Grep, Write, Edit) for inspection and documentation tasks.
- Sanitization: The instruction set includes a dedicated step to scan staged bundles for secrets and speech-derived content before finalizing the output.
Audit Metadata