AI Provider Concepts & Integration
This section provides comprehensive documentation on how ayaiay packs work with different AI providers and platforms. Each provider has unique concepts, capabilities, and integration patterns.
New: GitHub Copilot Integration Guide - Comprehensive documentation now available!
Overview
ayaiay packs are designed to work across multiple AI platforms. Understanding each provider's specific concepts helps you create more effective agents, prompts, and instructions.
Supported Providers
ayaiay OAuth (for Bots)
Focus: Secure machine-to-machine access from AI bots and agents to ayaiay APIs
Key Concepts:
- OAuth Clients: Per-publisher bot credentials managed in the dashboard
- Client Credentials Flow: Token issuance without interactive user login
- Scoped Tokens: Fine-grained permissions (for example
packs:read,skills:execute) - Discovery Endpoint: Standard metadata via
/.well-known/openid-configuration
Best for: AI bots, automation services, and server-to-server integrations
View ayaiay OAuth Documentation →
GitHub Copilot
Focus: Code-centric AI assistance with deep GitHub integration
Key Concepts:
- Agents: Custom AI assistants defined in
.github/copilot-instructions.md - Instructions: Behavior guidelines and context
- Tools: Built-in and custom functions agents can use
- Workspace Context: Automatic understanding of repository structure
Best for: Code generation, code review, testing, and development workflows
View GitHub Copilot Documentation →
Claude/Anthropic
Focus: Extended conversations with large context windows
Key Concepts:
- Projects: Organized workspaces with custom knowledge
- System Instructions: High-level behavior configuration
- MCP Servers: Model Context Protocol for extended capabilities
- Artifacts: Generated content that persists across conversations
Best for: Complex reasoning, large document analysis, and creative tasks
Google Gemini
Focus: Multimodal AI with native tool use
Key Concepts:
- Grounding: Connection to real-time information
- Extensions: Direct integrations with Google services
- Multimodal Input: Text, images, audio, and video understanding
- Function Calling: Structured tool invocation
Best for: Multimodal tasks, Google Workspace integration, and research
OpenAI
Focus: Versatile AI with extensive tool ecosystem
Key Concepts:
- Assistants: Persistent AI agents with state
- Functions: Custom tools and API integrations
- Threads: Conversation history management
- Retrieval: Built-in document search capabilities
Best for: Building AI applications, custom workflows, and API integrations
Cross-Platform Concepts
Common Artifacts
While each provider uses different terminology, ayaiay normalizes these concepts:
| ayaiay Concept | GitHub Copilot | Claude | Gemini | OpenAI |
|---|---|---|---|---|
| Agent | Custom Agent | Assistant with Project | Grounded Agent | Assistant |
| Instruction | Instructions file | System Instructions | System Instruction | Assistant Instructions |
| Prompt | Prompt Template | Prompt with Variables | Prompt Template | Message Template |
| Tool | Function | MCP Tool/Server | Function Declaration | Function |
| Context | Workspace Context | Project Knowledge | Grounding | Retrieval |
Agent Capabilities
Different providers offer different levels of capability:
| Capability | GitHub Copilot | Claude | Gemini | OpenAI |
|---|---|---|---|---|
| Code Generation | ⭐⭐⭐⭐⭐ | ⭐⭐⭐⭐ | ⭐⭐⭐ | ⭐⭐⭐⭐ |
| Large Context | ⭐⭐⭐ | ⭐⭐⭐⭐⭐ | ⭐⭐⭐⭐ | ⭐⭐⭐ |
| Multimodal | ⭐⭐ | ⭐⭐⭐ | ⭐⭐⭐⭐⭐ | ⭐⭐⭐⭐ |
| Tool Use | ⭐⭐⭐⭐ | ⭐⭐⭐⭐⭐ | ⭐⭐⭐⭐ | ⭐⭐⭐⭐⭐ |
| Real-time Data | ⭐⭐⭐ | ⭐⭐ | ⭐⭐⭐⭐⭐ | ⭐⭐⭐ |
Creating Cross-Platform Packs
When building ayaiay packs that work across providers:
- Use Standard Formats: Follow ayaiay manifest structure
- Document Provider-Specific Features: Note when features are provider-specific
- Test Across Platforms: Verify your pack works with multiple providers
- Provide Alternatives: Offer fallbacks for provider-specific capabilities
Next Steps
Choose your primary provider to learn more about their specific concepts and best practices:
- GitHub Copilot Documentation - For code-focused development
- Claude Documentation - For complex reasoning tasks
- Gemini Documentation - For multimodal applications
- OpenAI Documentation - For building custom AI applications
- ayaiay OAuth Documentation - For bot authentication and machine-to-machine API access