1# Agent Panel
  2
  3The Agent Panel provides you with a way to interact with LLMs.
  4You can use it for various tasks, such as generating code, asking questions about your code base, and general inquiries such as emails and documentation.
  5
  6To open the Agent Panel, use the `agent: new thread` action in [the Command Palette](../getting-started.md#command-palette) or click the ✨ (sparkles) icon in the status bar.
  7
  8If you're using the Agent Panel for the first time, you'll need to [configure at least one LLM provider](./custom-api-keys.md#providers).
  9
 10## Overview {#overview}
 11
 12After you've configured some LLM providers, you're ready to start working with the Agent Panel.
 13
 14Type at the message editor and hit `enter` to submit your prompt to the LLM.
 15If you need extra room to type, you can expand the message editor with {#kb agent::ExpandMessageEditor}.
 16
 17You should start to see the responses stream in with indications of which [tools](./tools.md) the AI is using to fulfill your prompt.
 18For example, if the AI chooses to perform an edit, you will see a card with the diff.
 19
 20### Editing Messages {#editing-messages}
 21
 22Any message that you send to the AI is editable.
 23You can click on the card that contains your message and re-submit it with an adjusted prompt and/or new pieces of context.
 24
 25### Checkpoints {#checkpoints}
 26
 27Every time the AI performs an edit, you should see a "Restore Checkpoint" button to the top of your message.
 28This allows you to return your code base to the state it was in prior to that message.
 29This is usually valuable if the AI's edit doesn't go in the right direction.
 30
 31### Navigating History {#navigating-history}
 32
 33To quickly navigate through recently opened threads, use the {#kb agent::ToggleNavigationMenu} binding, when focused on the panel's editor, or click the hamburger icon button at the top left of the panel to open the dropdown that shows you the six most recent interactions with the LLM.
 34
 35The items in this menu work similarly to tabs, and closing them from there doesn't delete the thread; just takes them out of the recent list.
 36
 37You can also view all historical conversations with the `View All` option from within the same menu or by reaching for the {#kb agent::OpenHistory} binding.
 38
 39### Following the Agent {#following-the-agent}
 40
 41Zed is built with collaboration natively integrated into the product.
 42This approach extends to collaboration with AI as well.
 43
 44As soon as you send a prompt to the Agent, click on the "crosshair" icon at the bottom left of the panel to follow along as it reads through your codebase and performs edits.
 45
 46### Get Notified {#get-notified}
 47
 48If you send a prompt to the Agent and then move elsewhere, putting Zed in the background, a notification will pop up at the top right of your monitor indicating that the Agent has completed its work.
 49
 50You can customize the notification behavior or turn it off entirely by using the `agent.notify_when_agent_waiting` key.
 51
 52### Reviewing Changes {#reviewing-changes}
 53
 54If you are using a profile that includes write tools, and the agent has made changes to your project, you'll notice the Agent Panel surfaces the fact that edits have been applied.
 55
 56You can click on the accordion bar that shows up right above the panel's editor see which files have been changed, or click `Review Changes` ({#kb agent::OpenAgentDiff}) to open a multi-buffer to review them.
 57Reviewing includes the option to accept or reject each edit, or accept or reject all edits.
 58
 59Diffs with changes also appear in individual buffers.
 60So, if your active tab had changes added by the AI, you'll see diffs with the same accept/reject controls as in the multi-buffer.
 61
 62## Adding Context {#adding-context}
 63
 64Although Zed's agent is very efficient at reading through your code base to autonomously pick up relevant files, directories, and other context, manually adding context is still usually encouraged as a way to speed up and improve the AI's response quality.
 65
 66If you have a tab open when triggering the Agent Panel, that tab will appear as a suggested context in form of a dashed button.
 67You can also add other forms of context, like files, rules, and directories, by either typing `@` or hitting the `+` icon button.
 68
 69You can even add previous threads as context with the `@thread` command, or by selecting "Start new from summary" option from the top-right menu in the agent panel to continue a longer conversation and keep it within the size of context window.
 70
 71Images are also supported, and pasting them over in the panel's editor works.
 72
 73### Token Usage {#token-usage}
 74
 75Zed surfaces how many tokens you are consuming for your currently active thread in the panel's toolbar.
 76Depending on how many pieces of context you add, your token consumption can grow rapidly.
 77
 78With that in mind, once you get close to the model's context window, we'll display a banner on the bottom of the message editor offering to start a new thread with the current one summarized and added as context.
 79You can also do this at any time with an ongoing thread via the "Agent Options" menu on the top right.
 80
 81## Changing Models {#changing-models}
 82
 83After you've configured your LLM providers—either via [a custom API key](./custom-api-keys.md) or through [Zed's hosted models](./models.md)—you can switch between them by clicking on the model selector on the message editor or by using the {#kb agent::ToggleModelSelector} keybinding.
 84
 85## Using Tools {#using-tools}
 86
 87The new Agent Panel supports tool calling, which enables agentic collaboration with AI.
 88Zed comes with [several built-in tools](./tools.md) that allow models to perform tasks such as searching through your codebase, editing files, running commands, and others.
 89
 90You can also extend the set of available tools via [MCP Servers](./mcp.md).
 91
 92### Profiles {#profiles}
 93
 94Profiles introduce a way to group tools.
 95Zed offers three built-in profiles and you can create as many custom ones as you want.
 96
 97#### Built-in Profiles {#built-in-profiles}
 98
 99- `Write`: A profile with tools to allow the LLM to write to your files and run terminal commands. This one essentially has all built-in tools turned on.
100- `Ask`: A profile with read-only tools. Best for asking questions about your code base without the fear of the agent making changes.
101- `Minimal`: A profile with no tools. Best for general conversations with the LLM where no knowledge of your code is necessary.
102
103You can explore the exact tools enabled in each profile by clicking on the profile selector button > `Configure Profiles…` > the one you want to check out.
104
105#### Custom Profiles {#custom-profiles}
106
107You may find yourself in a situation where the built-in profiles don't quite fit your specific needs.
108Zed's Agent Panel allows for building custom profiles.
109
110You can create new profile via the `Configure Profiles…` option in the profile selector.
111From here, you can choose to `Add New Profile` or fork an existing one with your choice of tools and a custom profile name.
112
113You can also override built-in profiles.
114With a built-in profile selected, in the profile selector, navigate to `Configure Tools`, and select the tools you'd like.
115
116Zed will store this profile in your settings using the same profile name as the default you overrode.
117
118All custom profiles can be edited via the UI or by hand under the `assistant.profiles` key in your `settings.json` file.
119
120### Model Support {#model-support}
121
122Tool calling needs to be individually supported by each model and model provider.
123Therefore, despite the presence of tools, some models may not have the ability to pick them up yet in Zed.
124You should see a "No tools" disabled button if you select a model that falls into this case.
125
126We want to support all of them, though!
127We may prioritize which ones to focus on based on popularity and user feedback, so feel free to help and contribute.
128
129All [Zed's hosted models](./models.md) support tool calling out-of-the-box.
130
131### MCP Servers {#mcp-servers}
132
133Similarly to the built-in tools, some models may not support all tools included in a given MCP Server.
134Zed's UI will inform about this via a warning icon that appears close to the model selector.
135
136## Text Threads {#text-threads}
137
138["Text threads"](./text-threads.md) present your conversation with the LLM in a different format—as raw text.
139With text threads, you have full control over the conversation data.
140You can remove and edit responses from the LLM, swap roles, and include more context earlier in the conversation.
141
142For users who have been with us for some time, you'll notice that text threads are our original assistant panel—users love it for the control it offers.
143We do not plan to deprecate text threads, but it should be noted that if you want the AI to write to your code base autonomously, that's only available in the newer, and now default, "Threads".
144
145### Text Thread History {#text-thread-history}
146
147Content from text thread are saved to your file system.
148Visit [the dedicated docs](./text-threads.md#history) for more info.
149
150## Errors and Debugging {#errors-and-debugging}
151
152In case of any error or strange LLM response behavior, the best way to help the Zed team debug is by reaching for the `agent: open thread as markdown` action and attaching that data as part of your issue on GitHub.
153
154This action exposes the entire thread in the form of Markdown and allows for deeper understanding of what each tool call was doing.
155
156You can also open threads as Markdown by clicking on the file icon button, to the right of the thumbs down button, when focused on the panel's editor.
157
158## Feedback {#feedback}
159
160Every change we make to Zed's system prompt and tool set, needs to be backed by an eval with good scores.
161
162Every time the LLM performs a weird change or investigates a certain topic in your codebase completely incorrectly, it's an indication that there's an improvement opportunity.
163
164> Note that rating responses will send your data related to that response to Zed's servers.
165> See [AI Improvement](./ai-improvement.md) and [Privacy and Security](./privacy-and-security.md) for more information about Zed's approach to AI improvement, privacy, and security.
166> **_If you don't want data persisted on Zed's servers, don't rate_**. We will not collect data for improving our Agentic offering without you explicitly rating responses.
167
168The best way you can help influence the next change to Zed's system prompt and tools is by rating the LLM's response via the thumbs up/down buttons at the end of every response.
169In case of a thumbs down, a new text area will show up where you can add more specifics about what happened.
170
171You can provide feedback on the thread at any point after the agent responds, and multiple times within the same thread.