Blint makes AI-coding
predictable and scalable.
Define your best design ideas as patterns, document the purpose behind your code, and enforce both consistently — across every file, every commit, every contributor.
What's included
The core idea
AI-coded projects need guardrails
to scale well.
Blint introduces controlled code — guardrails that enforce your design opinions automatically. Patterns define how code should be written. Documentation defines why it exists. Not thorough technical specification, but higher-level design principles and domain concepts — enforced consistently across every file, every commit, every contributor.
Keeping Control
Controlled code answers two questions: why was it written, and how should it be designed?
The why
Documentation
Business purpose and domain concepts. What does this service do? Why does this workflow exist? Project-specific context so your AI understands the system it's building, not just the file it's editing.
The how
Patterns
Design decisions and coding conventions. How should data be validated? How should errors be handled? Reusable ideas that keep code consistent as it scales.
Both live as markdown files in your repo — easy to read, edit, and version alongside your code.
Patterns
Share your best ideas between projects and with others
Sometimes it's good to be opinionated. A pattern is your take on how something should be done — a design principle with teeth. Each targets specific files via globs and gets enforced automatically.
Demo Driven Development
Build components in isolation with Storybook. Comprehensive demos as living documentation.
Torben SkoEnvironment Configuration
Fail fast with explicit values. No defaults, no fallbacks — validate all required variables at startup.
Torben SkoAsync Loading Component
Centralise loading, error, and retry UI for React Query. One component handles the state machine.
Torben SkoData Model Handling
Shared Zod schemas as the single source of truth. Validate at boundaries, infer types from schemas.
Torben SkoE2E Testing with Storybook
Page objects and component tests with Playwright. Composable, isolated test strategy.
Torben SkoManaging LocalStack
Treat LocalStack as disposable. Destroy-first deploys for AWS local development.
Torben SkoTailwind Culling Fix
Use @source directive to prevent class purging in shared monorepo components.
Torben SkoOrganising .env Files
Separate .env files per app and environment. No shared configs across stages.
Torben SkoNull vs Undefined
Treat them as semantically distinct. Null means explicitly empty; undefined means not set.
Torben SkoService Events
Lightweight DynamoDB-backed event logging for business visibility without a third-party platform.
Torben SkoLambda Tips
Health checks, debugging flags (verbose, dry), and version reporting on every Lambda.
Torben SkoAPI Mocking Pattern
Clean swap between real and mock APIs using React Context. No conditionals in component code.
Torben SkoAI-Managed Documentation
Embed markdown docs in your codebase. Let Claude maintain the content with real diagrams.
Torben SkoStrict Schemas
Typed schemas for all data structures. No loose JSON or untyped dictionaries.
Torben SkoColocation
Keep related code together. Tests, styles, and utilities next to the code they serve.
Torben SkoFail Fast
Validate preconditions early. Throw on invalid state rather than continuing silently.
Torben SkoA clear, spec-driven workflow
Plan work as specs, implement with guidance, review before merging
Every feature or fix starts as a specification document. Blint guides the process from draft through implementation to review — checking that the code follows your patterns before it ships.
Draft
Write the spec — describe what to build and why
Implement
Build the feature in an isolated environment
Check
Enforce patterns and review the changes
Merge
Ship with confidence
And there's no need to leave Blint to do the implementation — it has agentic coding built right in.
Built-in agentic coding
Claude Code, wrapped with a streamlined workflow
Blint wraps Claude Code directly into the app, so you get full agentic coding without context switching. Start a spec, and Blint opens a Claude session pre-loaded with your patterns, documentation, and an isolated environment.
Sessions scoped to spec branches — no cross-contamination
Patterns and docs fed as context automatically
Uses your existing Claude subscription — no extra API key
Isolated environments for parallel work
Each spec gets its own environment so parallel work doesn't collide
When you start a spec, Blint can set up an isolated environment — a git worktree, a full copy, or just the original folder. Multiple specs can run in parallel without stepping on each other.
Original folder
Work directly in your project. No isolation — best for quick fixes.
Git worktree
Clean checkout on a new branch. Main project stays untouched.
Full copy
Complete project copy with dependencies. Self-contained — no rebuild.
CLAUDE.md management
Keep your AI's context current without manual editing
Blint manages its own sections inside your CLAUDE.md — the file Claude Code reads for project guidance. Your existing content stays untouched. One click to sync the managed sections.
Patterns — routes Claude to the right pattern files based on glob matchers
Documentation — key points from your colocated docs, kept in sync automatically
Conventions — industry principles like DRY, KISS, Fail Fast, toggled on per project
Pointers — project-specific notes like "use AWS profile personal" or "run tests with --watch"
CLAUDE.md
Managed by Blint
Patterns
Documentation
Conventions
Pointers
Get started with Blint
Free for public patterns. Download the desktop app for macOS or Windows.