The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses git clone, git rev-parse, ls, find, and pygount via the terminal tool to analyze repository structure and history. These are standard operations for a documentation generation tool.
[EXTERNAL_DOWNLOADS]: The skill performs git clone --depth 50 <url> to a temporary directory when the user provides a GitHub URL. This is necessary for the skill's primary purpose of documenting remote repositories.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data by reading source code files from the target repository using read_file and search_files. It includes boundary markers in its output instructions (e.g., 'don't assume the reader has the source README') and operates with a limited capability set focused on file system reads and documentation writes. The risk is minimized by the skill's specific focus on structural analysis rather than executing the code it reads.
[PRIVILEGE_ESCALATION]: The skill uses mktemp -d and mkdir -p to manage output and temporary directories. No privilege escalation attempts (like sudo) were found.

code-wiki