Loading...
Product
Core Features
Top Extensions
Company
Subscribe to our newsletter.
Get product updates and news in your inbox. No spam.
Chat with your local Hermes Agent AI assistant directly from Raycast.
GET /v1/models and POST /v1/chat/completionsAPI_SERVER_KEY)Set the following environment variables:
export API_SERVER_ENABLED=true
export API_SERVER_KEY=your-secret-token-here
Or configure via CLI:
hermes config set api_server.enabled true
hermes config set api_server.key your-secret-token-here
The API server will hot-reload the config automatically.
Your token is the value you set for API_SERVER_KEY environment variable or in your Hermes configuration.
When you first run a command, Raycast will prompt for your API Endpoint and Token. The endpoint depends on where Hermes is running relative to Raycast:
Use when: Raycast and Hermes Agent are on the same computer.
| Setting | Value |
|---|---|
| API Endpoint | http://127.0.0.1:8642 |
This is the default - no configuration changes needed on Hermes.
Use when: Hermes runs on another computer on your home/office network.
| Setting | Value |
|---|---|
| API Endpoint | http://<hermes-machine-ip>:8642 |
Setup required on the Hermes machine:
Find the machine's local IP:
ipconfig getifaddr en0 # WiFi
# or
ipconfig getifaddr en1 # Ethernet
Set the API server host to 0.0.0.0:
export API_SERVER_HOST=0.0.0.0
Restart Hermes API server for changes to take effect.
Use the local IP as your endpoint, e.g., http://192.168.1.50:8642
⚠️ Security Warning: Binding to
0.0.0.0exposes the API server to your entire local network. Risks include:
- Anyone on the same WiFi can attempt connections
- Public WiFi = public exposure
- If port forwarding is enabled on your router, it could be internet-accessible
The token provides some protection, but Tailscale (Option C) is strongly recommended for accessing Hermes from other machines. Only use this option on trusted private networks.
Use when: You want secure access from anywhere - home, office, mobile, etc.
| Setting | Value |
|---|---|
| API Endpoint | https://<machine-name>.<tailnet>.ts.net |
Setup required on the Hermes machine:
Install Tailscale on both machines and sign in to the same account.
On the Hermes machine, set up Tailscale serve:
tailscale serve --bg 8642
Get your serve URL:
tailscale serve status
Output: https://machine-name.tailca3a37.ts.net
Use that URL as your API Endpoint.
Benefits:
| Method | Security | Works Remotely | Setup Complexity | Recommended |
|---|---|---|---|---|
| Local | High (localhost only) | No | None | ✅ Yes |
| Local Network | ⚠️ Low (LAN exposure) | No | Low | Only on trusted networks |
| Tailscale | High (encrypted, private) | Yes | Medium | ✅ Yes - best for remote |
Quick Q&A - type a question, get an answer. Supports passing a question as an argument for automation.
Full conversation interface with:
Reads your clipboard and lets you ask questions about it. Great for:
Select text in any app, then run this command to:
Tip: Assign a keyboard shortcut in Raycast preferences for quick access.
The API server endpoint isn't enabled. Set API_SERVER_ENABLED=true and restart Hermes.
Make sure Hermes API server is running:
hermes status
Verify your token matches the API_SERVER_KEY environment variable or Hermes config.
Thanks to @asaphko for the original extension icon and inspiration for the API Server Status and Open Webchat commands. This extension is inspired by the OpenClaw Raycast plugin.
Quick question to Hermes
Open a conversation with Hermes
Ask Hermes about your clipboard content
Summarize, explain, translate, or improve selected text
Check Hermes API server connection status
Open Hermes webchat in browser
ask-hermes: Send questions to Hermes and get answersprocess-text: Transform text (summarize, explain, fix grammar, etc.)check-status: Diagnose connection issues with the Hermes API serverGET /v1/models as the primary liveness check (OpenAI-compatible). /health is optional metadata./v1/models instead of always sending hermes-agent.