The Agent Skills Directory

[SAFE]: The skill's primary function is the management of architectural documentation and domain terminology. It does not include any commands for network access, sensitive file system operations (e.g., accessing .ssh or .aws credentials), or arbitrary command execution.
[SAFE]: No patterns of obfuscation (Base64, zero-width characters, homoglyphs) or remote code execution were found. The skill is entirely composed of instructions for document maintenance and user interaction.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted repository content, creating a surface for indirect prompt injection.
Ingestion points: Reads GLOSSARY.md, GLOSSARY-MAP.md, CONVENTIONS.md, and project source code.
Boundary markers: The agent is instructed to use specific structured templates (as defined in GLOSSARY-FORMAT.md and CONVENTIONS-FORMAT.md).
Capability inventory: Limited to reading and writing markdown documentation files.
Sanitization: No specific sanitization logic is present for data interpolated into the prompt. Despite this surface, the risk is assessed as safe because the agent lacks the high-privilege capabilities necessary to act on malicious instructions within those files, and the behavior is central to the skill's primary purpose.

model-domain