The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill instructions (Step 5) explicitly command the agent to 'Write a script or test that runs the real code, run it, and paste what happened.' This enables the generation and execution of arbitrary code based on the context of the repository being analyzed.\n- [COMMAND_EXECUTION]: The agent is instructed to use various CLI tools, including gitnexus (commands: analyze, detect-changes, impact), grep, and unslop. These tools are used to index the codebase and format the final report.\n- [INDIRECT_PROMPT_INJECTION]: The skill operates on untrusted data from external code repositories, creating an attack surface for indirect prompt injection.\n
Ingestion points: Untrusted data enters the agent context through PR diffs, commits, and source code files.\n
Boundary markers: The skill does not define specific boundaries or ignore-instructions for the code it processes.\n
Capability inventory: The agent possesses the capability to run shell commands and execute dynamically generated scripts.\n
Sanitization: There are no documented procedures for sanitizing the input data or validating the generated scripts before execution.

blast-radius