Stephen Collins.tech

Studying how codebases actually behave.

We build tools and run analyses at the intersection of software complexity, codebase health, and AI-generated code. This org is where that work lives in the open.

---

Recent OSS Analyses

Last updated: 2026-03-24 · Full analysis archive

Repository	Highest Risk Function	Score
colinhacks/zod	convertBaseSchema	21.3
FlowiseAI/Flowise	OpenAIAssistant	21.2
react-hook-form	createFormControl	20.6
tldraw/tldraw	Editor	19.3
styled-components	sanitizeCSS	15.4

---

What We Study

Area	Description
Structural complexity	The 20% of functions causing 80% of bugs, incidents, and slowdowns
AI-generated code	How Copilot, Claude, and coding agents change codebase structure over time
OSS health over time	Longitudinal complexity drift in major open source projects
Agentic codebases	What happens to a codebase when agents write most of it

---

Recent Writing

• What Happens When You Run Hotspots on 102,000 Functions — stress-testing against VS Code, finding an O(N³) bug
• AI Agents Can Pass Tests. They Still Can't Maintain Systems. — how AI tools excel at writing but struggle with maintenance
• AI Made Code Cheap. The Bottleneck Is Now Understanding Systems. — comprehension, not production, is now the engineering constraint

---

Open Artifacts

• hall-of-fame/ — Legendary gnarly functions from OSS history, analyzed
• ai-code-studies/ — Research into AI-generated code patterns and failure cases

---

The Tool

Hotspots finds the riskiest functions in any repo — the ones with high complexity and high churn that cause most of your bugs and slowdowns.

macOS

brew install Stephen-Collins-tech/tap/hotspots
hotspots analyze .

Linux

curl -fsSL https://raw.githubusercontent.com/Stephen-Collins-tech/hotspots/main/install.sh | sh
hotspots analyze .

---

Also in this org

• ts-validator — Rust-based static analysis for runtime type safety in TypeScript
• intent-kit — Intent-driven workflow library for LLMs

---

hotspots.dev · Blog · GitHub