The Agent Skills Directory

[SAFE]: The skill operates strictly within the local development environment, focusing on managing internal project documentation in the .codestable/architecture/ directory.
[COMMAND_EXECUTION]: Employs standard local CLI tools, specifically git log for change detection and a project-internal utility .codestable/tools/search-yaml.py for metadata indexing. These operations are transparently defined and aligned with the skill's primary function.
[PROMPT_INJECTION]: The skill processes untrusted input from the project's source code and documentation. To mitigate the risk of indirect prompt injection, it enforces 'evidence-based' rules requiring all architectural assertions to be anchored to specific file and line numbers (file:line). Additionally, all persistence operations (writing files) require explicit user review and approval, preventing automated malicious actions.

cs-arch