Overview
AgentBase is a configuration-driven AI Agent backend for secondary development. It is described as a production-grade AI Agent framework / LLM agent scaffold built on deepagents, LangChain, and LangGraph. The project allows users to assemble and run production-grade AI agents / intelligent agent systems from YAML configuration, without writing boilerplate.
agentbase provides YAML configuration, pluggable extension registries, component factories, a 21-command CLI, and a FastAPI service layer with 145 REST/WebSocket routes. It wires together infrastructure such as model configuration, prompt templates, user management, API key management, session management, memory management, knowledge base with RAG, document parsing, task queues, API security, tracing, and evaluation. Every component is swappable via a one-line config change.
Key features include config-driven agent assembly validated by agentbase doctor, a pluggable registry system with 30 extension registries, a full API server with 145 routes, a RAG knowledge base with 9 document formats and 3 chunking strategies, 48 built-in tools, 9 middleware, enterprise hardening, and 3,831 tests with 90% coverage. Defaults are local-dev; production compose refuses to start without AGENTBASE_API_KEY, YAML auth.api_key, or a JWT secret.
Pricing
Free
Detailed plans have not been confirmed in our catalog. Check the official website for current limits and billing terms.
Visit WebsitePrices and limits may change. Confirm the currency, billing period, seat minimum and usage caps on the official website.
Use Cases
- Teams use agentbase to assemble production-grade AI agents from YAML configuration without writing boilerplate.
- Teams use agentbase to swap databases, embedding models, queues, or tracers by changing one line of config.
- Teams use agentbase to expose a FastAPI service layer with 145 REST/WebSocket routes.
- Teams use agentbase to build RAG knowledge bases from 9 document formats with 3 chunking strategies and pgvector retrieval.
- Teams use agentbase to run async tasks through a queue and start a queue worker process.
- Teams use agentbase to evaluate agent suites and produce reports.
- Teams use agentbase to harden APIs with API key auth, JWT/RBAC, OAuth2, CORS, rate limiting, and request tracing.
- Teams use agentbase to manage model, prompt, user, API key, session, memory, and knowledge base infrastructure.
Features
Configure agent backends with YAML
Extend agents through pluggable registries
Access backend through FastAPI routes
Build RAG knowledge bases
Connect tools through MCP
Run queued agent workloads
Deploy the backend with Docker
Reviews
No reviews yet. Be the first to share your experience!