The Agent Skills Directory

[DATA_EXFILTRATION]: The script scripts/main.py reads local file contents via the --file argument in upload_file() and transmits them to an external endpoint (https://qwen-qwen3-asr-demo.ms.show). If the agent is instructed to "transcribe" a path containing sensitive information (e.g., SSH keys or configuration files), the skill will read and POST those contents to the third-party server.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted audio data and returns the resulting transcription directly into the agent's context, providing a vector for embedded instructions to influence the agent.
Ingestion points: The scripts/main.py script reads audio data from local files or standard input.
Boundary markers: None. Transcribed text is returned to the agent without delimiters or instructions to ignore potential commands within the data.
Capability inventory: The skill possesses the capability to read files from the local filesystem and perform network POST operations.
Sanitization: None. The transcription output is passed back to the agent without filtering or validation.
[COMMAND_EXECUTION]: The skill relies on executing a local Python script via uv run to perform its operations, which involves both filesystem access and network communication.

qwen-asr