> For the complete documentation index, see [llms.txt](https://docs.interactive.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.interactive.ai/agents/agents-overview.md). # Agents Overview An **Interactive Agent** is a configurable AI assistant that lives inside a project. You describe what it is, attach the behavioral building blocks it should follow, and deploy it as a running service reachable at its own URL. The agent is the **central artifact of the platform**, and most of what you create in a project, including prompts, glossaries, policies, and routines, exists to shape how your agents behave. Every agent is made of two main parts: * The **configuration** describes how the agent behaves: its identity, the system prompt that defines its role, the context it pulls from the project catalog, any extra rules, and the external services it connects to. * The **deployment** describes how the agent runs as a service: the runtime image, the public endpoint, environment variables, API keys, attached secrets, and an optional schedule. You build and manage both from a single agent view. The same configuration can be deployed more than once, and each deployment carries its own runtime settings. ### How Agents Relate to Your Project Agents belong to projects, and projects belong to your organization. An organization can hold many projects, and **each project can hold many agents**. Everything an agent draws on comes from the project it lives in. You create a component of the context, such as a policy, routine, or prompt, once on its own page within the project, and it becomes **reusable** across that whole project. From there, any agent in the same project can attach it and use it. These components stay within their project, so an agent built in one project cannot see the context, secrets, or traces of another. ### Lifecycle and statuses An agent becomes a running service once you deploy it. Every time you save a change to its configuration, the platform records a new version, so you always keep a full history you can compare against or roll back to if needed. Deploying a new version replaces the live one, and the version it replaced is marked as superseded. Agents and their versions carry a status that tells you where they are in this lifecycle: * **Draft** means the configuration has been saved but not yet deployed. The agent has no running instance. * **Running** means the agent is live and handling requests at its endpoint. * **Deactivated** means the agent has been taken out of service and is not currently running. * **Rollbackfailed** means an attempt to restore a previous version did not complete and needs attention. ### Working with an agent **All Agents** lists every agent in the project with its status and the time it was last updated, in either a list or a card layout. Each row gives you a shortcut to open a chat with the agent and to pin it, which places it in the Agents group in the sidebar so you can reach it in one click from anywhere in the project. Opening an agent takes you to its **detail view**, where you can edit its configuration, review its history, and test it with Open chat without leaving the platform. The **configuration** is shown two ways, switched with the Form and YAML tabs. Form mode lays the configuration out as the same sections you fill in when creating an agent, with the endpoint, runtime, and schedule settings grouped together under Deployment. YAML mode shows the raw configuration, which you can edit directly. Both reflect the same underlying data, so a change in one appears in the other. A panel on the left lists every saved version with its number, date, and status, and the header always shows which version you are viewing. Select any version to see its full configuration, and use **Compare versions** to put two revisions side by side and spot exactly what changed. To bring back an earlier revision, open it and use **Restore revision**. No rebuild is needed, since the platform redeploys the selected revision as it was. **Delete agent**, at the bottom of the view, removes the agent from the project. #### Editing an agent Editing an agent works the same way whether you are **refining** it for the first time or **updating** something already in production. You **change the components** you want in the UI and deploy, without rebuilding the agent or touching any code. Your earlier versions stay in the history, so you can always compare against them or roll back. When you **open a deployed agent**, a banner reminds you that you are editing the live agent, and the actions at the bottom of the view are worth knowing. **Save draft** creates a new pending version without touching the live agent. **Deploy configuration** applies your changes to the live agent right away. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.interactive.ai/agents/agents-overview.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.