{"skill":{"slug":"control-mirror","displayName":"control-mirror","summary":"Audits agent and system architecture through a control theory lens for stability, feedback, noise, delay, oscillation, error control, adaptive behavior, and...","tags":{"latest":"1.1.0"},"stats":{"comments":0,"downloads":133,"installsAllTime":0,"installsCurrent":0,"stars":1,"versions":3},"createdAt":1776756408845,"updatedAt":1777094806565},"latestVersion":{"version":"1.1.0","createdAt":1777093867564,"changelog":"# Changelog\n\nAll notable changes to `control-mirror` will be documented in this file.\n\n## [1.0.0] - 2026-04-25\n\n\n# Changelog\n\n## 2026-04-25\n\n### Added\n\n- Added a full Engineering Cybernetics review framework for system architecture audits.\n- Added control-system mapping for:\n  - reference input\n  - controller\n  - controlled object\n  - actuator\n  - sensor\n  - feedback\n  - noise\n  - delay\n  - error\n  - damping\n  - saturation\n  - stability boundary\n- Added a structured review workflow:\n  - define the control loop\n  - detect instability patterns\n  - measure real architectural strengths\n  - score control maturity\n  - recommend evolution steps\n- Added a 0-5 control maturity model:\n  - L0 Open loop\n  - L1 Human-corrected loop\n  - L2 Verified loop\n  - L3 Damped closed loop\n  - L4 Adaptive closed loop\n  - L5 Self-evolving controlled system\n- Added a general Control Scorecard covering:\n  - feedback completeness\n  - stability / convergence\n  - noise filtering\n  - delay handling\n  - error control\n  - damping / resource control\n  - observability\n  - adaptation\n  - safety boundary\n- Added domain-specific review guidance for:\n  - software platforms and microservices\n  - data pipelines and analytics systems\n  - workflow and operations systems\n  - agent / LLM systems\n  - organization and socio-technical systems\n- Added an AgentOS / Multi-Agent extension for reviewing:\n  - execution kernels\n  - model routing\n  - token control\n  - workflow gates\n  - memory pollution resistance\n  - feedback-to-policy loops\n  - safety boundaries\n- Added Darwin-style test prompts for future skill evaluation and regression testing.\n\n### Improved\n\n- Refocused the skill around the core ideas of Engineering Cybernetics rather than a single AgentOS-specific use case.\n- Improved the trigger description for better discovery in architecture review, control theory, feedback-loop, and multi-agent system scenarios.\n- Made the review output more actionable by requiring:\n  - control diagnosis\n  - consequences\n  - prioritized evolution actions\n  - verification criteria\n- Clarified that architectural strengths should only count when they improve controllability, stability, observability, or adaptation.\n- Added anti-patterns to prevent overly theoretical or generic architecture reviews.\n\n### Changed\n\n- Moved AgentOS-specific guidance into a dedicated extension section so the main skill remains broadly useful for general system architecture reviews.\n- Reorganized the skill into a more ClawHub-friendly structure:\n  - core lens\n  - cybernetics mapping\n  - review workflow\n  - scorecard\n  - output format\n  - domain-specific prompts\n  - AgentOS extension\n  - anti-patterns\n  - quality bar\n\n### Files\n\n- Updated `SKILL.md`\n- Added `references/agentos-review-rubric.md`\n- Added `test-prompts.md`","license":"MIT-0"},"metadata":null,"owner":{"handle":"xb19960921","userId":"s172f3jjxf40vh438bf24d3cj983g86d","displayName":"xb19960921","image":"https://avatars.githubusercontent.com/u/83778953?v=4"},"moderation":null}