Does this skill help with documentation?

Yes, it specifically identifies knowledge gaps during work and prompts the creation of subsystem specs or CLAUDE.md rules to ensure critical context isn't lost between sessions.

What is 'Compound Thinking' in this context?

Compound thinking is a methodology where every engineering task is used to improve the underlying system or process, ensuring that the next unit of work is inherited from a more efficient baseline.

How does this skill differ from a standard retrospective?

Unlike traditional retrospectives that happen after work is finished, this skill provides an active lens to identify, log, and capture improvements while you are actually performing the task.

How do I activate the improvement mindset?

You can activate this skill by asking Claude to 'work with improvement mindset', 'apply compound thinking', or 'set up an improvement cycle' before starting a task.

What are Subsystem Invariants?

Invariants are specific rules, conditions, or contracts within a specific part of your codebase that must always remain true to ensure stability and predictability.

Engineering Improvement Cycle

Name: Engineering Improvement Cycle
Author: mberto10

bymberto10

0•

Productivity & Workflow

Embeds a compounding improvement mindset directly into the engineering workflow to capture knowledge and eliminate friction in real-time.

The Engineering Improvement Cycle skill transforms standard development tasks into opportunities for continuous growth by applying a 'compound thinking' lens. Instead of simple task completion, it guides Claude to actively notice patterns, identify friction points, and encode subsystem knowledge during the execution phase. By grounding work in existing invariants and proactively documenting new learnings, this skill ensures that every engineering session contributes to a more efficient and robust codebase, creating a feedback loop where future tasks inherit the wisdom of previous ones.

Key Features

01Active pattern recognition for extracting reusable procedures and domain expertise

02Real-time friction logging to identify and solve workflow bottlenecks immediately

030 GitHub stars

04Subsystem-aware context loading to ensure work aligns with existing contracts and invariants

05Structured 'Before, During, and After' checklists to maintain high-quality engineering standards

06Automated encoding filters to determine which learnings should become rules, skills, or documentation

Use Cases

01Standardizing subsystem knowledge and invariants across a growing or complex codebase

02Reducing technical debt by proactively identifying and documenting 'why' specific architectural decisions were made

03Executing complex engineering tasks where past learnings need to be leveraged and new patterns identified

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add mberto10/mberto-compound improvement-cycle

For use in Claude.ai and ChatGPT

Download Skill