# Agent Editor Agent Builder allows you to see, open, edit, and test your agents. Once you build an worker using [Agent Builder](/toolhouse/toolhouse/quickstart.md), you will be able to change any aspect of your worker in the editor. Agent Editor gives you two modes depending on the kind of change you need to make. **Chat** is the best starting point for most edits. Use it when the issue isn't fully clear yet, or when you want to describe a change in plain language and let the editor figure out the implementation. You can point out unexpected behavior, ask why the worker did something, or describe what you want differently — the editor will update the prompt and integrations accordingly. > "The worker is sending the summary before it finishes reading all the files. Fix that." "Explain why the worker sends a Slack message even when there are no results." **Details** is for direct, precise edits. Use it when you know exactly what needs to change — a specific line in the system prompt, a constraint to add, an integration to remove. The Details tab gives you full access to every field without going through the chat interface. A good workflow is to use Chat for planning and exploration, then switch to Details to fine-tune. ## Editing your worker details manually The Details tab shows you an advanced view of every aspect of your worker. ### Agent Details You can change Title, Start message, and System prompt for your agent. You can also make this agent only visible to you, or change its custom URL. * **Title:** The worker's name. In email conversations, this will be the name that will appear in the **From** field. * **Start message:** When you use [Schedules](/toolhouse/toolhouse/schedules.md) or any other automated trigger, this will be the first message sent to your worker. * **System prompt:** This is the prompt created by Agent Builder when you use the chat tab. It contains all the necessary instructions (including guardrails and use of integrations) to make your worker perform its task. * **Agent ID:** The unique identifier of your worker. You cannot change this value. * **Visibility:** If the option is toggled, it will allow you to share your agent with others and to open up its inbox to everyone. Turn this off to disable access, including email access. When the agent is private, only you can use it. * **Custom URL:** Allows you to choose a friendly name for your chat interface and worker email address. ### Integrations You can control what individual actions the worker can perform. Each connected integration has multiple actions. For example, a Google Drive integration has the ability to search, move, list, and download files (and many others). Your worker likely won't need all of these integrations. Agent Editor is built to select the correct actions for your worker, but if you're editing your worker manually, you may need to add or change integrations from this section. ### Vibe Code The Vibe section contains a prompt to connect this worker to other services like Claude Code or Lovable. Use this when you want to build a custom interface, internal tool, or integration on top of your worker without writing the connection logic from scratch. ### Agent Files Agent files are knowledge files you can make available to your worker. They can contain guidelines, process descriptions, templates, and anything your worker needs to keep in mind in order to complete its task effectively. For example, if you are building a worker that generates email campaigns, you may want to provide an HTML file containing the template you always use. ### API Access Every worker can be called as an API. This means you can integrate your worker directly to your internal systems. --- # Agent Instructions: 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: ``` GET https://docs.toolhouse.ai/toolhouse/toolhouse/agent-editor.md?ask= ``` The question should be specific, self-contained, and written in natural language. 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.