Shows the logo of agenta

MIT license. Doc PRs welcome Contributors PyPI - Downloads Last Commit

Try Agenta Live Demo

Agenta is a platform for building production-grade LLM applications. It helps engineering and product teams create reliable LLM apps faster through integrated prompt management, evaluation, and observability.

Collaborate with Subject Matter Experts (SMEs) on prompt engineering and make sure nothing breaks in production.

• Interactive LLM Playground: Compare prompts side by side against your test cases
• Multi-Model Support: Experiment with 50+ LLM models or bring-your-own models
• Version Control: Version prompts and configurations with branching and environments
• Complex Configurations: Enable SMEs to collaborate on complex configuration schemas beyond simple prompts

Explore prompt management →

Evaluate your LLM applications systematically with both human and automated feedback.

• Flexible Testsets: Create testcases from production data, playground experiments, or upload CSVs
• Pre-built and Custom Evaluators: Use LLM-as-judge, one of our 20+ pre-built evaluators, or you custom evaluators
• UI and API Access: Run evaluations via UI (for SMEs) or programmatically (for engineers)
• Human Feedback Integration: Collect and incorporate expert annotations

Explore evaluation frameworks →

Get visibility into your LLM applications in production.

• Cost & Performance Tracking: Monitor spending, latency, and usage patterns
• LLM Tracing: Debug complex workflows with detailed traces
• Open Standards: OpenTelemetry native tracing compatible with OpenLLMetry, and OpenInference
• Integrations: Comes with pre-built integrations for most models and frameworks

Learn about observability →

Playground

Prompt Management

Evaluation

Observability

The easiest way to get started is through Agenta Cloud. Free tier available with no credit card required.

Try Agenta Live Demo

1. Clone Agenta:

git clone https://github.com/Agenta-AI/agenta && cd agenta

2. Copy configuration:

Before starting the services, create the environment file from the example:

cp hosting/docker-compose/oss/env.oss.gh.example hosting/docker-compose/oss/.env.oss.gh

3. Start Agenta services:

docker compose -f hosting/docker-compose/oss/docker-compose.gh.yml --env-file hosting/docker-compose/oss/.env.oss.gh --profile with-web --profile with-traefik up -d

4. Access Agenta at http://localhost.

For deploying on a remote host, or using different ports refers to our self-hosting and remote deployment documentation.

Find help, explore resources, or get involved:

• 📚 Documentation – Full guides and API reference
• 📋 Changelog – Track recent updates
• 💬 Slack Community – Ask questions and get support

We welcome contributions of all kinds — from filing issues and sharing ideas to improving the codebase.

• 🐛 Report bugs – Help us by reporting problems you encounter
• 💡 Share ideas and feedback – Suggest features or vote on ideas
• 🔧 Contribute to the codebase – Read the guide and open a pull request

Consider giving us a star! It helps us grow our community and gets Agenta in front of more developers.

Star us

All Contributors

Thanks goes to these wonderful people (emoji key):

This project follows the all-contributors specification. Contributions of any kind are welcome!

By default, Agenta automatically reports anonymized basic usage statistics. This helps us understand how Agenta is used and track its overall usage and growth. This data does not include any sensitive information. To disable anonymized telemetry set AGENTA_TELEMETRY_ENABLED to false in your .env file.