Getting Started
Relevant source files
The following files were used as context for generating this wiki page:
The bl1nk platform is an enterprise-grade IDE designed for creating, managing, and testing Claude AI skills. It operates as a desktop application that integrates a multi-provider AI environment with a specialized developer workflow focused on "Skills"—modular packages that extend AI capabilities with procedural knowledge and tool integrations.
The platform utilizes a monorepo architecture, leveraging Electron for the desktop client and a tRPC-based backend. Users can engage with the system through various chat modes, including Build, Ask, and Agent, each governed by specific system prompts that define the AI's interaction logic and technical constraints.
Sources: Comprehensive.md:7-12, apps/client/src/prompts/system_prompt.ts:326-340, skill/skill-creator/SKILL.md:12-16
Environment Setup
To begin development, the project requires a specific set of tools and a structured initialization process. The project uses pnpm as its primary workspace manager.
Prerequisites
- Node.js: Version >= 20.0.0
- Package Manager: pnpm (latest version)
- Version Control: Git
Sources: CONTRIBUTING.md:5-9
Initial Installation Steps
- Repository Setup: Clone the repository and install workspace-wide dependencies.
- Environment Configuration: Initialize environment variables using the provided example.
- Database Initialization: The project uses SQLite with Prisma for the server and Drizzle for the client. Migrations must be run to set up the local data stores.
- Service Startup: Development servers for both the client and server applications are started concurrently.
# Initial Setup
git clone https://github.com/bl1nk-org/claude-skill-builder.git
cd claude-skill-builder
pnpm install
cp .env.example .env
# Database Setup (Server)
cd apps/server
pnpm prisma:migrate
cd ../..
# Launch
pnpm dev
Sources: CONTRIBUTING.md:11-30
Architecture Overview
The platform follows a Monorepo structure divided into three main areas: applications, shared packages, and specialized AI resource directories.
Core Stack
| Component | Technology |
|---|---|
| Frontend | Next.js 15 + React 19 + Electron |
| Backend | Node.js + tRPC + Express |
| Database | SQLite + Prisma ORM / Drizzle |
| AI Integration | AWS Bedrock + OpenRouter |
| Package Management | pnpm monorepo |
Sources: Comprehensive.md:46-52, CONTRIBUTING.md:34-45
Data Flow Diagram
The following diagram illustrates the relationship between the Skill management system and the underlying database models.
Skill Development Workflow
A "Skill" is the primary unit of extension in the platform. It consists of instructions, specialized workflows, and bundled resources.
Skill Anatomy
A skill directory must follow a strict structure to be recognized by the platform:
- SKILL.md: Required file containing YAML metadata (name, description) and instructions.
- scripts/: Executable Python or Bash code for deterministic tasks.
- references/: Large documentation sets loaded into context only when needed.
- assets/: Visual or static files used in final outputs (templates, icons).
Sources: skill/skill-creator/SKILL.md:21-72
Creation Process Flow
The platform provides scripts to automate the lifecycle of a skill, from initialization to packaging for distribution.
Sources: skill/skill-creator/SKILL.md:124-180
AI Chat Modes
The IDE supports three distinct interaction modes, each configured with specific primary directives via system prompts.
| Mode | Purpose | Primary Directive |
|---|---|---|
| Build | Real-time code editing | Modify web applications and perform file operations using <bl1nk-write> tags. |
| Ask | Technical guidance | Provide conceptual explanations. ABSOLUTE PROHIBITION on generating code. |
| Agent | Research & Analysis | Gather API data and requirements before the coding phase begins. |
Sources: apps/client/src/prompts/system_prompt.ts:108-111, 237-251, 303-310
Interaction Sequence
This sequence shows how the "Build" mode handles code modifications while maintaining a live preview for the user.
Development Standards
To maintain high code quality and consistency across the monorepo, developers must adhere to the following standards:
- TypeScript: Strict mode is mandatory;
anytypes are prohibited. - Naming: PascalCase for React components, camelCase for utilities, and kebab-case for configuration files.
- State Management: Jotai is used for global client state, while TanStack Query handles server-side state synchronization.
- Styling: Tailwind CSS is the exclusive utility for layouts and design, supplemented by shadcn/ui components.
Sources: CONTRIBUTING.md:73-98, apps/client/src/prompts/system_prompt.ts:219-234
The bl1nk platform provides a structured environment for AI-driven development, combining robust monorepo architecture with strict skill-based extension patterns. By following these getting started guidelines, developers can effectively contribute to the IDE or build specialized AI agents within the platform's ecosystem.