# Documentation

> Documentation is an agent framework for multi-agent frameworks. Open-source toolkit for building customized AI agents from modular prompt and tool. It is available on API.

- **Kind:** Agent
- **Platforms:** API
- **Website:** https://agents-readthedocsio.readthedocs.io/en/latest/index.html
- **Categories:** [multi-agent frameworks](https://ai-powered.com/categories/multi-agent-frameworks)
- **Listing updated:** 2026-11-18

## About

The Agent project's documentation site describes a toolkit for assembling AI agents out of reusable parts. Its pages cover an autonomous agent section with methods such as from_config, act, step, observe and update_memory, along with a Standard Operation Procedure system for defining states, transitions and routing between them.

The modular design is the central idea: prompts are built from named components such as demonstration, chain-of-thought, rule, task and style pieces, and tools are added through components for knowledge base lookup, web search, web crawling, API calls and custom functions. Developers combine these to shape how an agent reasons and acts.

The documentation is aimed at developers who want to build agents with a configurable structure rather than a fixed chatbot. The site notes that the project is under active development, so interfaces may change.

## Features

- **Autonomous agent methods:** Agent classes expose methods such as from_config, act, step, observe and update_memory for running behaviour loops.
- **Standard Operation Procedure system:** States, transitions, routing and next-step logic let developers define agent workflows as explicit procedures.
- **Modular prompt components:** Prompts are assembled from nine named components covering demonstrations, reasoning, rules, tasks, style and input.
- **Tool components:** Pluggable tools cover extraction, knowledge base access, web search, web crawling, API calls and custom functions.
- **Environment and action definitions:** Agents can be given defined states, an environment and a set of actions.

## Good for

- Building a customized AI agent with specific prompt rules and tone
- Wiring web search or crawling into an agent's toolset
- Defining a multi-step agent workflow as a set of states and transitions
- Connecting an agent to a knowledge base or external API

## FAQ

### Is the project still in development?

The documentation says the project is under active development.

### What tool types can agents use?

The component list includes extraction, knowledge base, web search, web crawl, API, static and function components.

### What are the best alternatives to Documentation?

The closest alternatives to Documentation are Web, AgentForge, AIlice, Upsonic and llm_agents.

## Where this comes from

- Description: written from the project's website.

## Alternatives

- [Web](https://ai-powered.com/agents/web-3): Web is an agent framework for multi-agent frameworks.
- [AgentForge](https://ai-powered.com/agents/agentforge-3): AgentForge is an open-source agent framework for multi-agent frameworks.
- [AIlice](https://ai-powered.com/agents/ailice): AIlice is an open-source agent for multi-agent frameworks.
- [Upsonic](https://ai-powered.com/agents/upsonic): Upsonic is an open-source agent framework for multi-agent frameworks.
- [llm_agents](https://ai-powered.com/agents/llm-agents): llm_agents is an open-source agent framework for ai models and llm tooling.
- [ai-legion](https://ai-powered.com/agents/ai-legion): ai-legion is an open-source agent for multi-agent frameworks.
- [CowAgent](https://ai-powered.com/agents/cowagent): CowAgent is an open-source agent for personal assistants.
- [crewAI](https://ai-powered.com/agents/crewai): crewAI is an open-source agent framework for multi-agent frameworks.

---
Source: https://ai-powered.com/agents/documentation-3
