The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted input from conversation history to generate content for a local file, presenting a surface for indirect prompt injection. This is a low-risk concern inherent to the skill's primary function of glossary management.
Ingestion points: Reads from the current conversation history to scan for domain-related terms (SKILL.md).
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are used for the scanned content (SKILL.md).
Capability inventory: Includes file-write access to create and update UBIQUITOUS_LANGUAGE.md in the working directory (SKILL.md).
Sanitization: No specific sanitization or filtering of conversation content is mentioned prior to writing to the file (SKILL.md).
[SAFE]: The skill operates entirely locally, writing data to a specific file named UBIQUITOUS_LANGUAGE.md. It does not perform network operations, access sensitive system files, or execute external code.

ubiquitous-language