The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill directs the agent to install and use third-party, community-maintained packages udio_wrapper (Python) and udio-wrapper (Node.js/TypeScript). These packages are not officially affiliated with the Udio platform and originate from individual GitHub accounts (flowese and josephgodwinkimani).
[REMOTE_CODE_EXECUTION]: By instructing the installation of the aforementioned packages from public registries (PyPI, npm) or directly from Git repositories, the skill introduces a risk of remote code execution through unverified software dependencies.
[COMMAND_EXECUTION]: The skill requires the execution of shell commands for dependency management (pip install, npm install) and credential management (security on macOS). It also uses browser automation to interact with the Udio web interface.
[CREDENTIALS_UNSAFE]: The skill instructions guide the user to manually extract a session cookie (sb-api-auth-token) from their browser and provide it to the agent. While the skill includes documentation on secure storage (e.g., system keychain), the manual handling and hand-off of high-privilege authentication tokens represent a security risk.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources, creating a potential attack surface.
Ingestion points: Data is ingested from the Udio API and from browser snapshots of the udio.com platform (documented in SKILL.md and browser.md).
Boundary markers: Absent. There are no explicit instructions to the agent to treat external content from API responses or web pages as untrusted or to use delimiters.
Capability inventory: The skill has the ability to execute shell commands, perform network requests via API wrappers, automate browser actions, and write files to the ~/udio/ directory.
Sanitization: Absent. The instructions do not describe any validation or filtering of content received from Udio's servers before it is processed by the agent.

Udio