The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to use uvx showboat, which triggers the download and execution of the showboat package from the Python Package Index (PyPI) at runtime.
[COMMAND_EXECUTION]: The agent is directed to execute shell commands such as sed, grep, cat, find, and wc through the showboat exec wrapper. This allows the agent to read and display contents from the local file system.
[INDIRECT_PROMPT_INJECTION]: The skill's core function is to read and analyze source code files, creating an attack surface for indirect prompt injection if those files contain malicious instructions.
Ingestion points: The agent reads source code files from the local repository (referenced in Pass 1 and Pass 4).
Boundary markers: There are no specific delimiters or instructions provided to the agent to distinguish between code and embedded instructions.
Capability inventory: The skill allows for file reading (cat, sed, grep), file writing (showboat init), and external tool execution (uvx).
Sanitization: No explicit content sanitization or validation of the analyzed files is performed before processing.

walkthrough