The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill fetches content from external URLs and interacts with an external AI wizard. This introduces a surface for indirect prompt injection where malicious instructions embedded in web pages or AI responses could influence the agent's behavior. * Ingestion points: url parameter in collection endpoints and responses from the AI wizard. * Boundary markers: Absent. * Capability inventory: Command execution via curl, file system access for uploads, and resource management. * Sanitization: Not specified in the documentation.
[Data Exposure & Exfiltration] (LOW): The skill performs network requests to api.omnibox.pro, which is not a pre-approved whitelisted domain, and supports uploading local files to this external service. While this is the intended purpose of the skill, it creates a vector for data exfiltration if the agent is manipulated into uploading sensitive files.
[Command Execution] (SAFE): The skill uses curl to interact with the API. There is no evidence of arbitrary shell command injection or downloading and executing remote scripts via pipes.

omnibox-api