deepgram-python-text-to-speech

Pass

Audited by Gen Agent Trust Hub on Oct 9, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill instructions describe an environment that ingests untrusted text data for audio synthesis, creating a potential surface for prompt injection if the source text is attacker-controlled.
  • Ingestion points: The text parameter in client.speak.v1.audio.generate and the text field within SpeakV1Text used in WebSocket streaming.
  • Boundary markers: Absent; the provided code snippets do not include delimiters or specific instructions to treat input text as data rather than instructions.
  • Capability inventory: The code demonstrates capabilities to perform network requests (Deepgram API synthesis) and write files to the local filesystem (output.raw).
  • Sanitization: Absent; there is no mention of sanitizing, escaping, or validating the input text before it is sent to the synthesis engine.
  • [EXTERNAL_DOWNLOADS]: The documentation includes instructions to download and install additional components from an external source.
  • Evidence: The skill suggests running npx skills add deepgram/skills to obtain cross-language product knowledge and MCP setup. This targets the vendor's own namespace.
  • [COMMAND_EXECUTION]: The skill provides shell commands intended for execution to extend the agent's capabilities.
  • Evidence: The npx skills add deepgram/skills command is provided in a bash block for setup purposes.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 9, 2026, 10:49 AM