The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted issue data, creating an indirect prompt injection surface. * Ingestion points: Issue titles, descriptions, and comments are retrieved through commands like br show and br list. * Boundary markers: The skill documentation advises agents to use the --json flag to receive structured data, which helps in isolating content from instructions. * Capability inventory: The agent is authorized to execute the br CLI tool and perform git operations. * Sanitization: There is no mention of sanitizing or filtering instructions that might be embedded within the issue tracker data.
[EXTERNAL_DOWNLOADS]: The br upgrade command is included for tool self-updates. This is a standard administrative feature for maintaining the CLI utility and is considered a normal vendor function.

br