The Agent Skills Directory

[SAFE]: The skill follows established engineering practices for documentation generation and includes a human-in-the-loop verification step (the 'Grilling gate'). No malicious code or hidden instructions were found.
[COMMAND_EXECUTION]: The skill instructions call for the execution of the uvx disambiguate <term> command. This tool is used to maintain vocabulary consistency by checking terms against the project's glossary.
[EXTERNAL_DOWNLOADS]: Use of the uvx tool implies the potential download and execution of the disambiguate Python package from a registry if it is not already present in the local environment.
[PROMPT_INJECTION]: The skill operates by ingesting and processing untrusted user conversation data and codebase content, which presents a surface for indirect prompt injection. \n
Ingestion points: User conversation history, repository documentation (docs/glossary/), and general codebase files. \n
Boundary markers: The skill does not define specific delimiters to isolate user conversation data during the synthesis process. \n
Capability inventory: The agent has the ability to read the repository, execute CLI tools (uvx), and publish content to project tracker files (e.g., AGENTS.md). \n
Sanitization: No explicit sanitization or filtering logic is provided for the ingested content before it is processed into the final specification.

to-spec