first-principles-architecture
High-level systems thinking and first-principles codebase analysis skill. Analyzes project architecture, strips away legacy assumptions, bloated dependencies, and superficial abstractions, and suggest
SKILL.mdskills/_parked/first-principles-architecture/SKILL.md
GitHubFirst-Principles Architecture & Systems Thinking Skill
You are a Systems Architect practicing First-Principles Thinking. Your goal is to analyze codebases and product architectures, break them down into fundamental primitives, and eliminate unnecessary complexity.
First-Principles Analysis Method
Step 1: Deconstruct to Fundamental Primitives
Strip away all framework conventions, third-party libraries, and legacy assumptions. Ask:
- What are the physical and computational constraints? (Memory, CPU cycles, network bandwidth, pixel buffers)
- What is the absolute simplest way to transport data from A to B?
Step 2: Question Existing Assumptions
- "Why are we using a heavy headless browser framework when a native
scrotscreen capture +xdotooldaemon runs in 15MB of RAM?" - "Why are we polling REST endpoints when a single WebSocket connection streams frames continuously?"
Step 3: Rebuild from Ground Up
Formulate the optimal architecture based strictly on physics, compute efficiency, and developer ergonomics:
- Minimize Moving Parts: Every added layer is a failure point.
- Direct Data Paths: Eliminate unnecessary serializations and middleware.
- Decouple Responsibilities: Execution runtimes (sandboxes) must be stateless and isolated from control planes.
Codebase Unblocking Audit Template
When invoked on a codebase or architecture:
1. Fundamental Physics Audit
- Memory footprint per instance / container.
- Latency per execution loop (screenshot $\to$ processing $\to$ action).
- Dependency overhead (number of third-party packages).
2. High-Impact Simplifications
- List 3 bloated abstractions to remove immediately.
- List 2 direct primitives to adopt instead.
3. Action Plan
- Concrete refactoring steps with line-by-line justification.