The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to install the '@blackbox_ai/blackbox-cli' package via npm. This third-party software is used to communicate with the Blackbox AI service and manage coding tasks.\n- [COMMAND_EXECUTION]: The skill uses the 'terminal' tool to run the 'blackbox' command-line interface. This tool has the capability to modify files and execute coding workflows. The skill highlights the '--yolo' flag, which instructs the CLI to auto-approve all actions and model switches, effectively removing the human-in-the-loop requirement for operations.\n- [INDIRECT_PROMPT_INJECTION]: The skill provides examples for processing untrusted external content, such as reviewing code from a cloned GitHub repository or checking out a Pull Request (e.g., 'gh pr checkout 42 && blackbox --prompt "Review this PR..."'). This creates a surface where malicious instructions embedded in the external code could be executed or followed by the agent.\n
Ingestion points: Ingests code from external git repositories via 'git clone' and 'gh pr checkout' commands in SKILL.md.\n
Boundary markers: None identified; external code is processed by the AI without explicit delimiters or safety instructions to ignore embedded commands.\n
Capability inventory: The 'terminal' tool executes the 'blackbox' binary, which has broad permissions to edit files and manage project structures.\n
Sanitization: There is no evidence of sanitization or filtering of the external code before it is passed to the AI model for review.\n- [DYNAMIC_EXECUTION]: The CLI supports dynamic features including 'blackbox update', 'blackbox extensions', and 'blackbox mcp', which involve fetching and executing additional components or updates from remote sources at runtime.

blackbox