Set your API key
Open the Command Palette with Don’t have an API key yet? You can get one from any of these providers:
Ctrl+Shift+P (Windows/Linux) or Cmd+Shift+P (macOS), type Hermes: Set API Key, and press Enter. Paste your key into the prompt that appears — the input is masked so your key is never visible in plain text.Your API key is stored in VS Code Secret Storage, which encrypts keys at rest using the operating system’s credential store. The key is never written to
settings.json, never committed to source control, and never logged. You can update it at any time by running Hermes: Set API Key again.| Provider | Link |
|---|---|
| OpenRouter (access many models via one key) | openrouter.ai/keys |
| Anthropic (Claude models) | console.anthropic.com |
| OpenAI (GPT models) | platform.openai.com/api-keys |
Open the chat panel
Click the Hermes Agent icon in the Activity Bar — the vertical strip of icons on the far left of the VS Code window. The Hermes sidebar opens and the extension begins connecting to the agent in the background.Alternatively, open the Command Palette and run Hermes: Open Chat.
Wait for Ready
Watch the status pill at the top of the chat panel. It starts at Connecting…, moves to Initializing…, and finally turns green and reads Ready.On first run, initialization can take 1–3 minutes while Hermes loads its plugins and checks ACP dependencies. Subsequent starts are faster. If the status stays stuck on Connecting…, run Hermes: Detect Environment from the Command Palette (or click the search icon in the Hermes toolbar) to diagnose the issue.
Explore the Control Center
Open the embedded Hermes dashboard by clicking the dashboard icon in the Hermes toolbar, or run Hermes: Open Control Center from the Command Palette. The Control Center lets you select the active model, configure tool permissions, import or export profiles, and preview the system prompt — all without leaving VS Code.
Send your first message
Click the message input at the bottom of the chat panel, type a question, and press Enter to send. Use Shift+Enter to add a newline without sending.Hermes streams its reply into the panel as it generates — you will see the response appear character by character. Press Stop in the toolbar at any time to cancel a response mid-stream.A few prompts to try on your first conversation:
- “Explain this file and what it does.”
- “How do I add error handling to this function?”
- “What are the main risks in this code?”
- “Refactor this to use async/await.”
Pull in editor context
Hermes already has access to your workspace through the Editor Tools Bridge, but you can also send a specific selection directly into the chat input.Select any code in the editor, right-click, and choose Hermes: Insert Selection into Chat. The selected text is pasted into the message input so you can add a question before sending.You can also type
@ inside the chat input to open the workspace file picker and attach a specific file to your message.When you are ready to start a fresh conversation, click the + button in the Hermes toolbar or run Hermes: New Chat from the Command Palette. Each new chat is saved as a separate session tab — you can rename, delete, or switch between sessions at any time.What to try next
Now that your first conversation is working, explore the features that make Hermes most useful day-to-day.Chat
Learn about session management, context attachment,
@file references, and how to read file paths directly from replies.Editor Tools
Understand what the agent can see in your workspace and how to use CodeLens to trigger prompts from inside a file.
Models & Profiles
Switch between models and providers, compare per-token costs, and set up named agent configurations for quick switching.
Settings
Configure the Hermes binary path, permission mode, working directory, fallback model list, and more.