The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes a set of platform-specific CLI tools (prefixed with gbrain) to manage repository content. Specifically, it uses gbrain entity for named-entity resolution, gbrain search for semantic lookup, and gbrain get to read existing file contents. These operations are internal to the platform's repository management workflow.
[SAFE]: No malicious patterns such as remote code execution, data exfiltration, or obfuscation were detected. The skill's logic is focused on data integrity and organization within a trusted workspace environment.
[SAFE]: Although the skill reads existing repository content (which represents a potential surface for indirect prompt injection if that content were malicious), it does so within a controlled 'gate' process that uses delegated skills for actual write operations, minimizing the risk of unauthorized actions.

brain-ingest-gate