qwencloud-audio-tts
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/gossamer.pyusessubprocess.runto execute maintenance commands, including the installation of update-check tools vianpxand the execution of sibling update scripts. This functionality is scoped to the vendor's repository and local skill environment. - [EXTERNAL_DOWNLOADS]: The shared library
scripts/qwencloud_lib.pydownloads JSON configuration and model metadata from an Alibaba Cloud CDN (alioth-intl.alicdn.com). Additionally,scripts/tts.pyperforms downloads of synthesized audio files from vendor-managed storage. - [DYNAMIC_EXECUTION]: The update mechanism in
scripts/gossamer.pydynamically locates and executescheck_update.pyscripts from sibling skill directories within the agents' local environment. - [INDIRECT_PROMPT_INJECTION]: The skill processes arbitrary user-provided text for audio synthesis, which involves sending the data to external TTS services. Ingestion points: Text inputs provided by the user in synthesis requests through
scripts/tts.pyandscripts/tts_cosyvoice.py. Boundary markers: Not present; the scripts do not wrap the text in delimiters or provide explicit instructions to the API to ignore embedded commands. Capability inventory: Synthesizes audio via network calls to vendor APIs based on the provided text. Sanitization: The input text is passed directly to the synthesis API without content-based sanitization or filtering.
Audit Metadata