rethink

Synthesize accumulated observations into structured evolution proposals for human review.

Updated Apr 26, 2026
One-click install
npx skills add https://github.com/kinnerkarmanish/mak --skill rethink-kinnerkarmanish
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: rethink
Source: https://github.com/kinnerkarmanish/mak/tree/main/library/skills/learning/rethink
Command: npx skills add https://github.com/kinnerkarmanish/mak --skill rethink-kinnerkarmanish

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Synthesizes accumulated observations into actionable system evolution proposals, ensuring changes are reviewed before implementation and reducing ad-hoc fixes.

Core Features & Use Cases

  • 5-phase rethink workflow: triage, methodology updates, pattern detection, proposal generation, and human approval.
  • Triggers on observation accumulation in /remember and can generate proposals for specific categories.
  • Prevents auto-implementation, preserving human oversight.

Quick Start

Provide a category to rethink, e.g., /rethink 'process', to generate proposals for review.

Frequently Asked Questions about rethink

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate system evolution proposals from accumulated observations?

To generate system evolution proposals, trigger a five-phase workflow that synthesizes accumulated observations from /remember into concrete, reviewable proposals requiring human approval before implementation. You can target specific categories like /rethink 'process'.

What is the workflow for turning categorized observations into actionable system changes?

The workflow for turning observations into system changes involves five phases: triage, methodology updates, pattern detection, proposal generation, and human approval. It prevents auto-implementation to preserve human oversight and reduce ad-hoc fixes.

Can I configure the analysis depth and thresholds for pattern detection in my observations?

Yes, you can configure analysis depth and thresholds for pattern detection. The workflow applies configurable thresholds across categorized observations in /remember to synthesize structured evolution proposals.

How do I ensure human approval is enforced before system changes are implemented?

Human approval is enforced during the final phase of the workflow. The system generates concrete, reviewable proposals but prevents auto-implementation, ensuring all system changes require explicit human approval before proceeding.

Does this approach work for synthesizing observations across specific categories?

Yes, it works for specific categories. You can provide a category to rethink, such as /rethink 'process', to generate targeted proposals. The workflow triggers on observation accumulation in /remember and processes categorized data.

What are the limitations of using automated proposal generation for system evolution?

The main limitation is that generated proposals cannot auto-implement changes; they strictly require human approval. Additionally, the workflow relies on accumulated observations in /remember, meaning insufficient or poorly categorized data may limit proposal quality.