docs: instructions for OpenCode

docs/getting-started.md

@@ -172,6 +172,31 @@ Create or edit `~/.qwen/settings.json`:
 
 Replace the placeholder API key with the providers you use (Gemini, OpenAI, OpenRouter, etc.).
 
+**For OpenCode CLI:**
+Edit `~/.config/opencode/opencode.json`:
+
+```json
+{
+  "$schema": "https://opencode.ai/config.json",
+  "mcp": {
+    "zen": {
+      "type": "local",
+      "command": [
+        "/path/to/zen-mcp-server/.zen_venv/bin/python",
+        "/path/to/zen-mcp-server/server.py"
+      ],
+      "cwd": "/path/to/zen-mcp-server",
+      "enabled": true,
+      "environment": {
+        "GEMINI_API_KEY": "your_api_key_here"
+      }
+    }
+  }
+}
+```
+
+Add any other API keys you rely on (`OPENAI_API_KEY`, `OPENROUTER_API_KEY`, etc.).
+
 #### IDE Clients (Cursor & VS Code)
 
 Zen works in GUI IDEs that speak MCP. The configuration mirrors the CLI examples above—point the client at the `uvx` launcher and set any required environment variables.
@@ -293,6 +318,11 @@ CUSTOM_MODEL_NAME=llama3.2                   # Default model name
 2. Run `qwen mcp list --scope user` and confirm `zen` shows `CONNECTED`.
 3. Try: `"/mcp"` to inspect available tools or `"Use zen to analyze this repo"`.
 
+### For OpenCode CLI:
+1. Restart OpenCode (or run `OpenCode: Reload Config`).
+2. Open **Settings › Tools › MCP** and confirm `zen` is enabled.
+3. Start a new chat and try: `"Use zen to list available models"`.
+
 ### For Codex CLI:
 1. Restart Codex CLI if running
 2. Open a new conversation