The Agent Skills Directory

[SAFE]: The skill's primary function is to provide a structured procedure for refactoring and modularizing AI agent skills. It facilitates the extraction of reusable workflows while maintaining the original functionality through an orchestrator pattern.
[SAFE]: The skill uses established, internal platform tools like workflow-skill and callstack-simulation to perform its tasks. These are intended for structural analysis and skill generation within the platform's ecosystem.
[SAFE]: The procedure includes multiple safety checkpoints, specifically requiring 'explicit approval of the whole current plan' from the user before any file modifications or creations occur (Step 7 and Step 8).
[SAFE]: The skill promotes the principle of least privilege by advocating for 'fresh-context agent/subagent' boundaries, which isolate tasks from unrelated history and use bounded handoff packets (Step 4 and Step 9).
[SAFE]: The skill includes verification and validation steps to ensure that the refactoring preserves behavior, maintains correct call graphs, and does not alter unrelated files (Step 10).

decompose-skill