The Agent Skills Directory

[SAFE]: The skill extracts class and function names from the local codebase to identify naming mismatches. This access is limited to read operations for documentation purposes and does not involve exfiltration or unauthorized access to sensitive files.- [PROMPT_INJECTION]: The skill analyzes conversation history and source code, creating a surface for indirect prompt injection. Ingestion points include domain stories and codebase identifiers processed during glossary generation. No explicit boundary markers or input sanitization steps are defined in the instructions, though the skill's capabilities are limited to writing a non-executable Markdown file.

ubiquitous-language