To support you at every stage, we created the Warden Community Agents repository: a curated collection of open-source agent examples and starter templates.
👉 Explore the repository: community-agents
This repository serves two goals at once: onboarding new builders and providing a shared place to showcase community-built agents.
Below, you’ll learn how each example can help you on your path from the first prototype to a production-ready agent.
🛠 Repository Overview: From Templates to Production-Ready Agents
In the repository, you’ll find three main areas:
👉 agents/: A collection of agent templates and fully working examples.
👉 docs/: Documentation covering LangGraph basics, setup, and deployment.
👉 Community Agents and Tools: A curated list where you can propose and showcase your own agent or tooling.
Each agent lives in its own directory under agents/ and is fully self-contained, with its own dependencies, configuration, and build setup.
💡 Most examples are written in TypeScript, with one Python example available as well.
💡 Using LangGraph is a qualification requirement for the incentive programme, so all examples are LangGraph-based agents. You can learn about other technical requirements here: How to Build Agents for the Incentive Programme: Requirements & Guidelines.
The agents overviews below are ordered from simple starter templates to more advanced, production-style agents, so you can choose what best matches your experience level.
LangGraph Quick Start Templates
To help you get up and running as fast as possible, the repository includes two LangGraph quick start templates:
👉 langgraph-quick-start (TypeScript)
👉 langgraph-quick-start-py (Python)
Each template implements a single-node chatbot that receives a user message, calls an OpenAI model, and returns a response.
🗎 To dive deeper, check out the accompanying guides:
These examples focus on the fundamentals: project structure, configuration, and deploying a minimal LangGraph agent. If you’d prefer to skip the basics and explore real-world agent logic right away, start with more advanced examples listed below.
Weather Agent
The Weather Agent is a beginner-friendly example and the recommended starting point for new agent developers:
This agent answers questions about weather: it fetches and processes data from WeatherAPI and then returns user-friendly results generated with OpenAI.
🗎 To better understand the example, check out these step-by-step guides:
The Weather Agent bridges the gap between “hello world” examples and more advanced systems. Because of its clarity and scope, it’s the best reference for builders preparing their first submission to the incentive programme.
CoinGecko Agent
The CoinGecko Agent moves into more complex territory, focusing on detailed analysis of digital assets and cryptocurrency market:
The agent retrieves pricing and market metrics from the CoinGecko API, analyzes asset performance and risks, and delivers comparative analysis for up to two tokens, generating results with OpenAI. Reasoning is implemented using the Schema-Guided Reasoning (SGR) approach, with a clearly defined five-step workflow:
Validate the request
Extract token information
Fetch market data
Validate the data
Analyze and summarize results
🗎 An overview of the example is available in the CoinGecko Agent README.
This example highlights patterns essential for production-grade agents: structured data handling, multi-step reasoning, and careful management of external dependencies. It’s especially useful for builders interested in financial data, DeFi tooling, or market intelligence use cases.
⭐ The CoinGecko Agent is live on Warden, where you can log in and test it directly.
Portfolio Analysis Agent
The Portfolio Analysis Agent is the most advanced example in the repository and demonstrates how to synthesize multiple data sources into higher-level insights:
It integrates the CoinGecko API and the Alchemy API to analyze cryptocurrency portfolios across EVM and Solana wallets. Using OpenAI and Schema-Guided Reasoning (SGR), the agent produces comprehensive portfolio reports through a structured workflow:
Parse the request and extract time periods and wallet addresses
Fetch the portfolio data (balances and historical prices)
Analyze the portfolio
Analyze the performance over time
Generate a detailed report
🗎 To explore the underlying logic, refer to this documentation:
If you’re aiming to submit a more sophisticated agent to the incentive programme, this example is a strong reference for overall architecture, reasoning flow, and scope.
⭐ The Portfolio Analysis Agent is live on Warden, where you can log in and test it directly.
🚀 What’s Next
The community-agents repository is designed to evolve alongside the Warden ecosystem. Very soon, we’ll release Warden Studio, a platform for publishing agents on Warden in a few clicks.
Meanwhile, you can explore the examples, build your own agents, and submit them to the team. Whether you’re experimenting with a LangGraph template or refining a production-ready agent, you’ll find support and feedback in the #developers channel on the Warden Discord.