The Agent Skills Directory

[SAFE]: The skill operates as a high-level conceptual framework for software architecture analysis. It contains no executable scripts (other than a trivial Python test witness), no external dependencies, and no remote data fetching operations.
[SAFE]: The instructions for utilizing the 'ledger' CLI tool are part of the skill's intended workflow for managing local documentation and do not involve high-risk command execution, privilege escalation, or unauthorized persistence.
[INDIRECT_PROMPT_INJECTION]: The skill processes repository content to perform architectural analysis. While this constitutes an ingestion surface for untrusted data, the instructions provide a rigid formal framework using Category Theory constructs and specific reporting templates. These act as constraints on the agent's reasoning process, significantly minimizing the risk of the agent obeying malicious instructions embedded in project code or comments.

universalist