Learn to Build Agents

Step-by-step guides to create powerful AI agents in multiple formats

Claude.md

Start with text

Code

LangChain/SDK

n8n

Visual workflow

Tutorial

Video walkthrough

Getting Started with Claude.md

Create a claude.md File

The easiest way to build an agent is to create a plain text file with agent instructions.

# My Research Agent
## Purpose
Analyze research papers and summarize findings.

## Instructions
1. Read the provided paper
2. Extract key findings
3. Generate executive summary

Structure: title, purpose, instructions, capabilities, examples

Define Your Agent's Behavior

Specify what your agent should do, what information it needs, and how it should respond.

## Capabilities

- Web search and analysis

- Document summarization

- Code generation

## Input Format

User provides: topic, document link, or code snippet

## Output Format

Structured JSON or markdown response

Upload to AgentStack

Share your agent with the community and enable others to use it instantly.

Start Uploading

Advanced Formats

n8n Workflows

Build complex automation workflows with visual node-based programming. Perfect for integration-heavy agents.

→ No coding required
→ 400+ integrations
→ Scheduled or webhook triggers

View n8n Examples

Python/LangChain

Build sophisticated agents using Python with LangChain, OpenAI SDK, or Anthropic SDK.

→ Full control & flexibility
→ Memory & context management
→ Custom tools & plugins

from langchain import OpenAI
agent = initialize_agent(...)
agent.run(prompt)

Best Practices

Clear Instructions

Write explicit, detailed instructions. Ambiguity leads to poor results.

Examples Matter

Include input/output examples. Show the agent what success looks like.

Test Thoroughly

Test with edge cases and unexpected inputs before uploading.

Document Dependencies

List all required APIs, credentials, or external tools.

Version Your Agents

Use semantic versioning (1.0.0) and describe changes in updates.

Collect Feedback

Monitor ratings and comments. Iterate based on user feedback.

Common Agent Patterns

Research Agent

Searches web, reads documents, compiles findings

SearchAnalysisSummarization

Code Generator

Writes, debugs, and explains code based on requirements

CodingDebuggingExplanation

Data Analyst

Processes data, creates visualizations, generates reports

DataAnalyticsReporting

Customer Support Bot

Answers questions, troubleshoots, escalates to humans

SupportFAQEscalation

Content Creator

Generates articles, social posts, email campaigns

WritingMarketingContent

Task Manager

Creates tasks, schedules meetings, sends notifications

AutomationSchedulingIntegration

Ready to Build?

Create your agent and share it with thousands of developers

Upload Your Agent