deaf-first-platform

360Magicians - AI Agent Platform

Intelligent automation and assistance agents for the MBTQ Universe.

Overview

360Magicians is a comprehensive AI agent platform providing intelligent automation, document processing, and assistance for deaf-first services. It supports multiple AI models and provides extensive capabilities for agent management, workflow orchestration, and knowledge management.

API Endpoints

Base URL: https://api.mbtquniverse.com/ai

Note: The endpoints below show full paths for clarity. In the OpenAPI specification, these are defined as relative paths with the base URL specified in the servers section. Path parameters use {paramName} syntax in the OpenAPI spec but are shown as :paramName here for readability.

Agents

• POST /ai/agents - Create agent
• GET /ai/agents - List agents
• GET /ai/agents/:agentId - Get agent
• PATCH /ai/agents/:agentId - Update agent config
• DELETE /ai/agents/:agentId - Delete agent
• POST /ai/agents/:agentId/clone - Clone agent

Execution and Runs

• POST /ai/agents/:agentId/execute - One-off execution with input
• POST /ai/agents/:agentId/runs - Start a long-running run
• GET /ai/runs - List runs
• GET /ai/runs/:runId - Run status and output stream pointer
• POST /ai/runs/:runId/cancel - Cancel run
• GET /ai/runs/:runId/logs - Execution logs
• GET /ai/runs/:runId/artifacts - Generated files or outputs

Tasks and Queue

• POST /ai/tasks - Enqueue task for any agent
• GET /ai/tasks - List tasks with filters
• GET /ai/tasks/:taskId - Task status
• POST /ai/tasks/:taskId/cancel - Cancel task
• POST /ai/tasks/:taskId/retry - Retry failed task

Tools and Capabilities

• GET /ai/tools - List available tools
• POST /ai/tools - Register custom tool
• GET /ai/tools/:toolId - Tool details
• PATCH /ai/tools/:toolId - Update tool
• DELETE /ai/tools/:toolId - Remove tool
• POST /ai/agents/:agentId/tools/:toolId/enable - Enable tool for agent
• POST /ai/agents/:agentId/tools/:toolId/disable - Disable tool for agent

Models and Routing

• GET /ai/models - List supported models
• PATCH /ai/agents/:agentId/model - Set primary model
• POST /ai/agents/:agentId/model/fallbacks - Set fallbacks or routing rules

Memory and Context

• GET /ai/agents/:agentId/memory - Fetch memory items
• POST /ai/agents/:agentId/memory - Upsert memory
• DELETE /ai/agents/:agentId/memory/:memoryId - Delete memory
• POST /ai/agents/:agentId/context - Attach ephemeral context for next run

Files, Knowledge, and Embeddings

• POST /ai/agents/:agentId/files - Upload knowledge file
• GET /ai/agents/:agentId/files - List files
• DELETE /ai/agents/:agentId/files/:fileId - Remove file
• POST /ai/agents/:agentId/ingest - Start ingestion and embedding
• GET /ai/agents/:agentId/ingest/:jobId - Ingestion status
• POST /ai/search - RAG style query across agent knowledge

Workflows and Graphs

• POST /ai/workflows - Create workflow DAG
• GET /ai/workflows - List workflows
• GET /ai/workflows/:workflowId - Workflow details
• POST /ai/workflows/:workflowId/execute - Run workflow
• GET /ai/workflow-runs/:runId - Workflow run status

Scheduling

• POST /ai/agents/:agentId/schedules - Create schedule
• GET /ai/agents/:agentId/schedules - List schedules
• PATCH /ai/schedules/:scheduleId - Update
• DELETE /ai/schedules/:scheduleId - Delete

Webhooks and Events

• POST /ai/webhooks - Register webhook
• GET /ai/webhooks - List webhooks
• DELETE /ai/webhooks/:webhookId - Delete
• POST /ai/events/test - Send test event
• GET /ai/events - Event log

Secrets and Configuration

• GET /ai/secrets - List names
• POST /ai/secrets - Set or rotate secret
• DELETE /ai/secrets/:name - Delete secret
• GET /ai/config - Current global config and limits

Analytics and Audit

• GET /ai/metrics - Usage metrics per agent
• GET /ai/metrics/costs - Model and tool cost breakdown
• GET /ai/audit - Audit log of actions and runs

Health and Operations

• GET /ai/health - Liveness/readiness
• GET /ai/version - Build and model registry versions

Features

• Multiple AI model support (GPT-4, Gemini, Claude, and more)
• Agent lifecycle management
• Workflow orchestration with DAG support
• Vector embeddings and RAG (Retrieval-Augmented Generation)
• Task queuing and scheduling
• Tool extensibility
• Memory management
• File ingestion and processing
• Webhook integration
• Cost tracking and analytics

Environment Variables

# 360Magicians Configuration
MAGICIANS_AI_KEY=your_ai_key_here
MAGICIANS_MODEL=gpt-4
MAGICIANS_OPENAI_API_KEY=your_openai_key
MAGICIANS_ANTHROPIC_API_KEY=your_anthropic_key
MAGICIANS_GOOGLE_API_KEY=your_google_key
MAGICIANS_MAX_TOKENS=4096
MAGICIANS_TEMPERATURE=0.7
MAGICIANS_VECTOR_DB_URL=your_vector_db_url
MAGICIANS_EMBEDDING_MODEL=text-embedding-ada-002

Supported AI Models

360Magicians supports multiple AI model providers:

• OpenAI: GPT-4, GPT-4 Turbo, GPT-3.5 Turbo
• Anthropic: Claude 3 Opus, Claude 3 Sonnet, Claude 3 Haiku
• Google: Gemini Pro, Gemini Ultra
• Open Source: LLaMA, Mistral, and other open-source models

AI SDK Integration

The platform uses the AI SDK for unified model access and streaming support. It integrates with:

• OpenAI SDK
• Anthropic SDK
• Google AI SDK
• TensorFlow.js for client-side processing
• Hugging Face Transformers for specialized accessibility models

Integration with Deaf Research

360Magicians is designed to integrate with:

• Deaf research databases
• Sign language models
• Accessibility research repositories
• ASL recognition systems

OpenAPI Specification

See openapi/openapi.yaml for the complete API specification.

Related Repositories

• pinkycollie/pinksync - Accessibility engine
• pinkycollie/deafauth-ecosystem - Authentication system
• pinkycollie/fibonrose - Blockchain layer
• pinkycollie/pinkflow - Hub pipeline integrator

This site is open source. Improve this page.