Pre-built Agent Library
Relevant source files
The following files were used as context for generating this wiki page:
The Pre-built Agent Library is a core component of the bl1nk platform, providing a centralized repository of over 500 specialized AI agent templates. These templates are designed to transform the general-purpose Claude AI into a specialized assistant equipped with domain-specific procedural knowledge and instructions for various professional fields.
The library is organized into categorical directories within the agents/ root, facilitating easy discovery and management of agents ranging from software development and technical writing to business consulting and creative arts. This modular structure supports the platform's goal of improving AI skill development by providing ready-to-use personas that follow standardized JSON schemas for system roles and metadata.
Sources: Comprehensive.md:7-10, CONTRIBUTING.md:12-21, scripts/generate-docs.js:37-43
Library Structure and Categorization
The library is undergoing a reorganization to improve accessibility, moving from a flat structure of 501 unorganized files to a categorized directory system. Each agent is represented by a JSON template containing its unique identifier, author information, metadata tags, and a detailed system role.
Agent Categories
| Category | Description | Estimated Count |
|---|---|---|
| Development | Coding, debugging, and testing specialized agents | 100+ |
| Writing | Content creation, translation, and technical writing agents | 80+ |
| Business | Marketing, consulting, and project management agents | 60+ |
| Education | Tutoring and academic assistance agents | 50+ |
| Creative | Art, design, and multimedia conceptual agents | 40+ |
Sources: Comprehensive.md:195-207, CONTRIBUTING.md:12-21
Directory Organization
The diagram shows the hierarchical organization of agents into domain-specific subdirectories to facilitate the organize:agents migration command.
Sources: Comprehensive.md:197-207, CONTRIBUTING.md:16
Agent Schema and Configuration
Agents are defined using a standardized JSON schema. This schema ensures consistency across the library and allows the platform to programmatically load agent personas into the system prompt.
JSON Structure
{
"author": "Author Name",
"identifier": "unique-agent-id",
"meta": {
"title": "Agent Title",
"description": "Agent description",
"tags": ["category", "tags"]
},
"systemRole": "Detailed system role instructions..."
}
Sources: CONTRIBUTING.md:100-111
Prompt Integration Logic
Pre-built agents are integrated into the application via the system prompt construction logic. The platform utilizes different prompt modes (Build, Ask, Agent) to define the AI's behavior based on the selected agent's context.
The prompt construction logic combines the selected agent's role with global technical rules defined in AI_RULES.md.
Sources: apps/client/src/prompts/system_prompt.ts:286-310
Validation and Health Checks
To maintain the integrity of the 500+ agent templates, the platform includes automated validation scripts. These scripts ensure that every agent adheres to the required schema and that metadata is complete for indexing.
- pnpm validate:agents: Executes the agent validator script to check JSON integrity and mandatory fields (identifier, systemRole).
- node scripts/health-check.js: Performs a high-level project health check, which includes verifying the completeness of the agent library.
- organize:agents: A specialized script used to migrate agents from the root directory into their respective categorical folders.
Sources: CONTRIBUTING.md:52-58, Comprehensive.md:209-211
Related Systems
The Pre-built Agent Library works in tandem with the Skill Platform and the tRPC API to deliver specialized capabilities. While agents provide the persona and system role, Skills provide the modular packages of specialized knowledge and tools that those agents can utilize.
Comparison Table
| Feature | Agents | Skills |
|---|---|---|
| Purpose | Defines persona and behavior | Defines procedural knowledge and tools |
| Format | JSON Templates | SKILL.md + Resources (scripts/assets) |
| Scope | Domain-specific instructions | Multi-step workflows and API integrations |
| Location | agents/ directory | skill/ directory |
Sources: CONTRIBUTING.md:12-21, scripts/generate-docs.js:37-43, Comprehensive.md:23-26
Conclusion
The Pre-built Agent Library serves as the foundation for the bl1nk platform's specialized AI capabilities. By offering a categorized and validated collection of over 500 agents, the project enables developers to quickly deploy AI assistants tailored to specific professional domains, ensuring high-quality interactions through standardized system roles and comprehensive metadata.