The Agent Skills Directory

[COMMAND_EXECUTION]: The documentation references installation commands for the 'tilth' MCP server (tilth install <host>). These are presented as manual setup instructions for the user/environment and are not automated by the skill itself.
[PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection (Category 8) because its primary function is to read and display external file content.
Ingestion points: tilth_read retrieves content from arbitrary files in the repository.
Boundary markers: The tool uses structured headers (e.g., # path (lines, tokens) [mode]) and line numbering to delimit file content from agent instructions.
Capability inventory: The skill allows reading files (tilth_read), listing files (tilth_files), and checking dependencies (tilth_deps).
Sanitization: No explicit sanitization of file content is described; however, the use of structural markers and AST-based outlining reduces the risk of the agent misinterpreting content as commands.

cheez-read