ucdaw
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to translate subjective natural language feedback ('ear notes') from users into specific technical actions within a DAW, creating an attack surface where data influences control flow.
- Ingestion points: Natural language strings such as 'drums a touch loud' or 'can't hear the bass' are processed to determine numeric dB adjustments.
- Boundary markers: The instructions do not define specific delimiters or guards to isolate user-provided text from the agent's logic.
- Capability inventory: The skill can modify channel faders, input trims, track configurations, and import audio files from the local filesystem (e.g.,
~/Music/stems). - Sanitization: No evidence of input validation or sanitization for these natural language inputs is provided in the skill instructions.
- [COMMAND_EXECUTION]: The skill operates by executing a local CLI tool (
ucdaw) to perform tasks such as network discovery, reading DAW state, and manipulating parameters via shell commands. - [EXTERNAL_DOWNLOADS]: The documentation includes multiple references to external audio engineering resources and industry blogs (such as Sound On Sound, iZotope, and PC Audio Labs) to provide technical context for mixing workflows.
Audit Metadata