The Agent Skills Directory

[SAFE]: The skill is purely instructional and documentation-focused. It defines a workflow for creating and maintaining specific markdown files (SOURCE_OF_TRUTH.md, ARCHITECTURE_ROADMAP.md, TODO_WORKFLOW.md, and CLAUDE.md) as a system design methodology.
[SAFE]: There are no executable scripts, network requests, or third-party package dependencies included in the skill. All recommended actions involve creating or auditing markdown documentation.
[SAFE]: The audit mode is a read-only verification process that identifies drift between code and documentation and reports findings with file and line references, which is a standard development practice.
[SAFE]: The skill uses natural instructional language and does not contain patterns indicative of prompt injection, obfuscation, or data exfiltration. The mention of CLAUDE.md is a configuration pattern for the execution environment and does not introduce security risks.

architect