Frontend developer specialising in React and TypeScript, with a focus on developer tooling and open-source utilities. I build things that help teams write cleaner code, catch technical debt early, and ship more consistent interfaces - and I'm currently building Dynvo end to end, from the analysis engine to the product.

A live, feature-level map of your codebase — built for engineering teams and the AI agents working alongside them.

Dynvo turns months of git history into a structured map of your product. It detects the features in your code and the user flows that run through them, scores each one for health, and pins down exactly where technical debt and regression risk actually live - no Jira archaeology, no manual tagging, no annotations to maintain. One cold scan, two audiences:

For engineers - a "Signal Room" dashboard with per-feature health, coverage gaps, hotspots, change-impact (blast radius), and bus-factor risk. PR comments that show what a diff actually touches before you merge, plus weekly Slack digests of where the codebase is drifting.

For AI agents — an MCP server exposing 13 read-only tools that hand Cursor, Claude Code, and Cline a structured map of your product plus live production signal — so your agent edits with real context instead of grepping blind.

Runtime, on the map - Sentry errors and PostHog usage are attributed back onto the exact feature and flow they belong to. Activity and risk land where the work happens, not in a separate dashboard.

• Code-grounded, LLM-assisted feature & user-flow detection - monorepo / workspace aware
• Per-feature health: bug-fix ratio, churn, coverage, ownership & bus factor, change impact
• Native MCP for AI agents — your product map becomes first-class agent context
• Privacy by design — Standard, Private (customer-managed keys), and self-hosted Sovereign modes; source code is never stored at rest, on any tier
• Open-source CLI (MIT) + hosted platform — unlimited developers on every tier, pay for repos not seats

# 1 · Scan a repo locally with the open-source engine
pip install dynvo
dynvo scan-v2 ~/my-project --llm

# 2 · Give your AI agent the map (Cursor / Claude Code / Cline)
npx -y mcp-remote https://app.dynvo.ai/api/mcp

Tech behind it  

Lighthouse for design consistency.

A CLI tool that crawls websites using a real browser and audits them against design system best practices. Unlike static CSS analysers, it captures computed styles, JavaScript-injected values, and runtime-resolved CSS custom properties to reflect what users actually see.

What it checks:

• Typography — font families, sizes, and line-height consistency
• Color — unique palette size, shade clustering via delta-E, WCAG AA contrast validation
• Spacing — grid system adherence (4 px / 8 px), outlier detection
• Components — touch target sizes, button variants, border-radius systems, z-index organisation
• Outputs a structured JSON report ready for CI pipelines

npx design-auditor https://yoursite.com

CSV exports that handle pagination automatically.

A React hooks library that makes large CSV exports painless. Pass it a paginated data source and it handles the batching logic, so UI code stays clean.

Lightweight URL state for react-router-dom.

A minimal React hook that makes reading and writing URL query parameters as ergonomic as useState, with full compatibility with react-router-dom.

Open to collaboration on developer tooling, React libraries, and design-system automation.