The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands to establish links and project structures. Evidence includes the use of ln -s, cmd //c mklink //J, mkdir -p, and git rev-parse across macOS, Linux, and Windows environments to automate environment setup.- [EXTERNAL_DOWNLOADS]: The skill performs remote repository cloning based on user input. Evidence includes the execution of git clone if a git URL is provided by the user during the wiki discovery workflow.- [PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by processing external wiki content. 1. Ingestion points: The agent reads .wiki/CONTRIBUTING.md and scans wiki directories for .md files in Step 4. 2. Boundary markers: Absent; the skill incorporates discovered file paths and descriptions directly into AGENTS.md. 3. Capability inventory: Includes file writing to AGENTS.md, CLAUDE.md, and settings.json, and shell command execution for directory creation and repository cloning. 4. Sanitization: Absent; the skill relies on external file content to guide its configuration logic.

wiki-connect