Neura Market
Agenta logo

Agenta

Free

Open-source LLMOps platform for prompt management, LLM evaluation, and observability. Build, evaluate, and monitor production-grade LLM applications. [#opensource](https://github.com/agenta-ai/agenta)

FreeFree tier
Type
Open Source
Company
Agenta
Visit Website GitHub (40 stars)

About Agenta

Agenta is an open-source workspace for building, evaluating, and monitoring AI agents. Users can create agents through natural language chat, iterate with feedback, and share agents with their team. It supports version control for prompts, skills, and tools, provides tracing for every run, includes human-in-the-loop approval, and offers scheduling and event-triggered automation. Agents can be built with multiple model providers (Anthropic, OpenAI, Ollama, xAI, OpenRouter) and run locally or in the cloud. The platform also includes evaluations, playground for side-by-side comparison, and role-based access control. Pricing includes a free hobby tier, paid cloud plans, and an open-source self-hosted option.

Key Features

Build agents through natural language chat
Improve agents with feedback and iteration
Version control for prompts, skills, and tools
Trace every agent run with cost and step details
Human-in-the-loop approval for consequential actions
Schedule and event-triggered agent runs
Supports multiple model providers (Anthropic, OpenAI, Ollama, xAI, OpenRouter)
Open-source with self-hosting option
Evaluations and side-by-side playground comparison
Role-based access control and SSO (paid plans)

Pros & Cons

Pros
  • Open-source with a free forever tier (Hobby) for up to 2 team members
  • Very flexible – supports multiple LLM providers, harnesses, and runtimes
  • Built-in versioning and traceability for every agent run
  • Human-in-the-loop approval prevents unintended actions
  • Active community and regular updates visible on changelog and roadmap
  • Supports scheduling and event triggers for automated workflows
Cons
  • Free tier limited to 5,000 agent runs per month and 2 team members
  • Advanced features (SSO, audit logs, custom retention) require higher-tier plans or enterprise
  • Self-hosted setup may require technical expertise (Docker Compose)
  • Relatively new platform – ecosystem and integrations still growing

Best For

Code review agents that review pull requests and leave inline commentsCustomer support agents that query knowledge base and respondSales agents that research companies and draft outreachCompany operations agents that monitor dashboards and generate reportsInternal knowledge management agents that update documentation from data sources

FAQ

Is there a free tier?
Yes, Agenta offers a free Hobby plan that includes unlimited projects, agents, workflows, and up to 2 team members, with 5,000 agent runs per month and 1-week trace data retention.
Can I self-host Agenta?
Yes, Agenta is open-source and can be self-hosted using Docker Compose. The self-hosted version includes all core features with unlimited users, agents, and workflows.
Which AI models are supported?
Agenta supports multiple providers including Anthropic (Claude), OpenAI, Ollama, xAI (Grok), and OpenRouter. You can also bring your own gateway.
What is a 'harness' in Agenta?
A harness is the runtime environment that executes your agent. Supported harnesses include Claude Code, pi.dev Codex, and upcoming support for OpenCode, Hermes, and others. You can swap harness without changing your agent configuration.
Does Agenta support scheduling and triggers?
Yes, Pro and higher plans include unlimited schedules and event triggers. You can set agents to run on a recurring schedule (e.g., every Monday at 9:00) or trigger them from events like pull requests.