The Agent Skills Directory

[SAFE]: The skill implements a structured documentation workflow with strong constraints against code generation and execution. It explicitly forbids the use of code blocks or pseudocode in the output, which mitigates risks associated with malicious code injection or generation.
[SAFE]: While the skill involves reconnaissance of the local codebase (Phase 2), including reading build configurations and authentication middleware, this access is restricted to the agent's local environment for the purpose of architectural consistency. No network operations or data exfiltration patterns were detected.
[SAFE]: The instructions provide explicit 'Anti-slop rules' that require all design decisions to be traced back to the parent specification, acting as a functional guardrail against hallucinations or unauthorized design deviations.
[SAFE]: No obfuscation, prompt injection vectors, or suspicious dependency management patterns were identified in the skill instructions.

design-doc