Documentation
Everything you need to build, deploy, and scale production AI agents with AgentiCraft.
Quick Start
uv add agenticraft-foundation agenticraft-llm agenticraft-typesPackage Documentation
agenticraft-foundation
v0.1.0Formal algebra, CSP process calculus, topology analysis, and mathematical foundations for multi-agent systems.
- Formal verification with CSP algebra
- Network topology and Laplacian analysis
- Stochastic service mesh primitives
- MathJax-rendered proofs
agenticraft-llm
v0.1.0Production-grade LLM provider abstraction with cost-aware routing, key rotation, circuit breakers, and an OpenAI-compatible gateway.
- 14 LLM providers with unified API
- Cost-aware routing with Thompson sampling
- Multi-key rotation (5 strategies)
- Circuit breaker and rate limiting
agenticraft-types
v0.1.0Shared type definitions, Pydantic models, Protocol classes, and error hierarchy used across all AgentiCraft packages.
- Pydantic v2 request/response models
- Protocol-based duck typing interfaces
- 11-class error hierarchy
- Configuration schemas
Guides
Getting Started
Install all three packages and build your first agent
Architecture Overview
How Foundation, LLM, and Types fit together
Multi-Agent Patterns
Coordination, consensus, and distributed patterns
Production Deployment
Checklists, monitoring, and scaling for production
Security Best Practices
Authentication, encryption, and secure agent communication
Migration Guide
Migrate from LangChain, CrewAI, or custom frameworks