Agenta
: How to use it, features, and the business problems it solves
Add bookmark
What is Agenta?
An open-source (MIT) workspace for building and running AI agents. Build agents by chatting, connect them to your apps, and run them in the background with approval flows, tracing, and version history. Positioned as "the open-source workspace for your agents," it supports 1,000+ integrations (MCP/Composio) and is model-agnostic. Self-hosting is free, with a cloud version available.
Business problems it solves
About "Agenta"
What is Agenta
Agenta is an open-source (MIT) workspace for building and running AI agents: build agents by chatting, connect them to your apps, and run them in the background with approval flows, tracing, and version history. Under the banner "The open-source workspace for your agents," teams build agents together and improve them through feedback. Self-hosting is free, and a cloud version is also available.
Agenta has shifted its positioning from "prompt-management, evaluation-centric LLMOps" to a "workspace for building and running agents." Build agents by chatting, run them in the background on schedules or events, control which actions run automatically vs. require approval, trace model calls and track cost, keep version history, and reuse skills. It handles 1,000+ integrations (MCP/Composio) and harnesses like Claude Code and Codex.
How to use
Build agents by chatting, then connect and run them.
Build an agent
Instruct via chat to build an agent for your goal.
Connect
Link your own tools via 1,000+ integrations (MCP/Composio).
Run and approve
Run on schedules/events, and require approval for sensitive actions.
Improve and share
Review traces and history, improve, and share with the team.
Self-hosting is free under the MIT license (Docker Compose/Helm). The cloud Hobby tier is free with limits such as number of runs. Model token costs are borne on each provider's side.
Key features
01Chat-based agent building and sharing
Create agents by instructing via chat, and share and improve them with the team. It includes a shared-file workspace that you and your agents use, plus reusable skills.
02Background runs, approvals, tracing
Supports background runs triggered by schedules or events, and permission control over which actions run automatically vs. require approval. It also traces model calls, tracks requests, tokens, latency, and estimated cost, and keeps version history.
03Integrations, models, open source
Supports 1,000+ integrations (Gmail, Slack, Notion, GitHub, etc.) via MCP servers or Composio. It is agnostic to major providers such as OpenAI, Claude, Gemini, and Mistral, and can also handle harnesses like Claude Code and Codex. Under the MIT license, you can self-host with the same setup as the cloud.
Pricing
Self-hosting is free; the cloud tiers below are monthly (no annual billing is advertised).
| Plan | Monthly | Highlights |
|---|---|---|
| Hobby | $0 | 2 members, 5,000 runs/month, 1-week trace retention |
| Pro | $29 | Unlimited members, 10,000 runs/month, 1-month retention |
| Business | $299 | RBAC, SSO, SOC 2 Type II, 3-month retention |
| Enterprise | Contact us | Audit logs, dedicated deployment, etc. |
Self-hosting (open source) is free with unlimited users/projects. Model token costs apply separately on each provider's side. Prices are in USD and subject to change; check the official site for the latest (as of August 2026).
How it differs from other AI agent/LLM platforms
Agenta's strength is that you can build agents by chatting, run them in the background with approvals, tracing, and history, and self-host it as open source. Unlike app-building-centric platforms, it differs in operations (approvals, execution, observability) and in being open source and self-hostable.
| Aspect | Agenta | Dify | Traditional manual work |
|---|---|---|---|
| Focus | Agent building/running workspace | LLM app dev platform | Manual management |
| Strength | Approvals/tracing, history, OSS self-host | RAG/workflow building | Flexible but ad hoc |
| Models | Agnostic (major providers) | Various models | — |
LLM app platform Dify and traditional manual management are the comparison points. If you want to run everything from building agents to operating them with approvals, execution, and observability — while self-hosting as open source — Agenta is a strong choice.
The information on this page is based on the official website as of August 2026. Please check the official site for the latest specs and pricing.
