The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run to coordinate tasks across multiple sibling skills, including fetcher, memory, ingest-youtube, and agent-inbox. These calls are implemented using list-based arguments to prevent shell injection and are core to the skill's coordination functionality.
[EXTERNAL_DOWNLOADS]: The skill facilitates the fetching of content from arbitrary remote URLs as part of its failure-recovery logic. It leverages multiple strategies and external tools to ensure resilient data ingestion.
[PROMPT_INJECTION]: As a fetcher utility, the skill ingests untrusted data from URLs and manifest files, representing an indirect prompt injection surface. The skill manages the fetching workflow and hands off content for downstream processing, relying on the agent's overall architecture for content sanitization.
[SAFE]: No malicious patterns, such as hardcoded credentials, obfuscation, or unauthorized persistence mechanisms, were found. The skill maintains a clear separation of concerns and uses a human-in-the-loop approach for sensitive recovery actions.

debug-fetcher