Supercharge Your Agents: Build AI Workflows in Minutes with Fast-Agent

Unlock the power of sophisticated AI agents and workflows with fast-agent, the first framework offering complete MCP feature support, including sampling. Ready to build effective agents using Anthropic and OpenAI models? This guide shows you how.

Why Choose Fast-Agent?

• Rapid Development: Create agents and workflows in minutes.
• MCP Native: Leverages MCP (Model-Controller-Presenter) architecture directly.
• Multi-Modal Support: Handles images and PDFs with ease.
• Simple Syntax: Focus on prompts and MCP Servers, not boilerplate code.
• Easy Testing: Passthrough and playback LLMs streamline Python glue-code development.

Kickstart Your Agent Development

Ready to jump in? Here's how to get started:

1. Install: uv pip install fast-agent-mcp
2. Setup: fast-agent setup (creates example agent and config files)
3. Run: uv run agent.py
4. Specify Model: uv run agent.py --model=o3-mini.low
5. Explore: fast-agent quickstart workflow (for building effective agents examples)

Pro-Tip: Windows users may need configuration changes for Filesystem and Docker MCP Servers. Details are in the configuration files.

Building Basic Agents: It's Easier Than You Think

Defining an agent is incredibly straightforward with fast-agent. Just use the @fast.agent decorator!

@ fast. agent (
 instruction = "Given an object, respond only with an estimate of its size."
)

Now, interact with your agent:

async with fast. run () as agent:
 moon_size = await agent ( "the moon")
 print ( moon_size)

Or, engage in interactive chat:

async with fast. run () as agent:
 await agent. interactive ()

Combining Agents with MCP Servers

The real power of fast-agent comes from combining agents into workflows using MCP Servers. This lets you create complex chains of actions without tedious coding.

• Use fast-agent quickstart workflow to generate example workflows.
• fast-agent searches for configuration files in the current and parent directories.

@ fast. agent (
 "url_fetcher",
 "Given a URL, provide a complete and comprehensive summary",
 servers = [ "fetch"], # Name of an MCP Server defined in fastagent.config.yaml
)

@ fast. agent (
 "social_media",
 """
 Write a 280 character social media post for any given text.
 Respond only with the post, never use hashtags.
 """,
)

@ fast. chain (
 name = "post_writer",
 sequence = [ "url_fetcher", "social_media"],
)

Mastering Workflows: Chain, Parallel, and More

Fast-Agent offers several powerful workflow types to orchestrate your agents.

• Chain: Execute agents sequentially. Great for tasks that require a specific order of operations.
• Parallel: Run agents simultaneously for faster processing or ensembling diverse LLM outputs.
• Human Input: Agents can request help from a real person to handle ambiguous situations or complex decisions.
• Evaluator-Optimizer: Refine content iteratively, using one agent to generate and another to evaluate quality.
• Router: Route messages to the most appropriate agent based on content analysis.
• Orchestrator: Decompose complex tasks into sub-tasks and distribute them among available agents.

Deep Dive: Agent Features and Customization

Calling Agents:

• Flexibly call agents using names or default configurations.
• Use send or prompt functions for interactive sessions.

Defining Agents:

• Customize agents with names, instructions, MCP servers, models, and more.
• Control chat history, request parameters, and human input capabilities.

Multimodal Magic: Handling Images and PDFs

Fast-Agent excels at handling multimodal data. Add resources to prompts using the inbuilt prompt-server or MCP Types directly.

summary: str = await agent. with_resource (
 "Summarise this PDF please",
 "mcp_server",
 "resource://fast-agent/sample.pdf",
)

MCP Prompts and Sampling: Fine-Tuning for Success

• Use apply_prompt(name, arguments) for MCP Prompts.
• Configure sampling LLMs per client/server pair in fastagent.config.yaml.

Ready to Revolutionize Your AI Development?

Fast-agent is your key to unlocking rapid, efficient, and sophisticated AI agent development. Build multi-modal workflows, leverage powerful features, and create intelligent solutions in minutes. Start building effective agents today!