Add capabilities negotiation flow

Changes:
- Server sends 'request_capabilities' after handshake
- Client responds with 'capabilities_response' containing supported features
- Server computes intersection of account + client capabilities
- Effective capabilities used for message validation
- Client metadata stored per-connection
- Updated tests for capabilities negotiation
- Updated CLIENT_SPEC.md documentation

Protocol messages added:
- request_capabilities (server → client)
- capabilities_response (client → server)

ConnectedClient fields added:
- clientCapabilities?: ChannelCapabilities
- clientMetadata?: Record<string, unknown>
- capabilitiesReady: boolean

New methods:
- getEffectiveCapabilities(client) - returns intersection
- clientSupportsCapability(client, capability) - check support

docs/CLIENT_SPEC.md

@@ -30,6 +30,25 @@ ws.on('message', (data) => {
     case 'handshake':
       console.log('Server capabilities:', message.capabilities);
       break;
+    case 'request_capabilities':
+      // Send client capabilities in response
+      ws.send(JSON.stringify({
+        type: 'capabilities_response',
+        capabilities: {
+          text: true,
+          images: true,
+          files: false,
+          voice: false,
+          reactions: false,
+          threads: true,
+          typing: true
+        },
+        metadata: {
+          clientName: 'My App',
+          version: '1.0.0'
+        }
+      }));
+      break;
     case 'message':
       console.log('Bot says:', message.content.text);
       break;
@@ -63,7 +82,25 @@ def on_message(ws, message):
     print(f"Received: {data}")
     
     if data['type'] == 'handshake':
-        print(f"Capabilities: {data['capabilities']}")
+        print(f"Server capabilities: {data['capabilities']}")
+    elif data['type'] == 'request_capabilities':
+        # Send client capabilities in response
+        ws.send(json.dumps({
+            "type": "capabilities_response",
+            "capabilities": {
+                "text": True,
+                "images": True,
+                "files": False,
+                "voice": False,
+                "reactions": False,
+                "threads": True,
+                "typing": True
+            },
+            "metadata": {
+                "clientName": "My Python App",
+                "version": "1.0.0"
+            }
+        }))
     elif data['type'] == 'message':
         print(f"Bot: {data['content']['text']}")
 
@@ -151,12 +188,12 @@ Sent once when connection is established.
   "accountId": "default",
   "capabilities": {
     "text": true,
-    "images": false,
+    "images": true,
-    "files": false,
+    "files": true,
     "voice": false,
     "reactions": false,
-    "threads": false,
+    "threads": true,
-    "typing": false
+    "typing": true
   },
   "metadata": {
     "serverName": "Production",
@@ -167,8 +204,10 @@ Sent once when connection is established.
 ```
 
 **Fields:**
-- `capabilities` - What this account supports
+- `capabilities` - **Maximum capabilities allowed by server** (account config)
-- `metadata` - Custom metadata from config
+- `metadata` - Custom metadata from server config
+
+**Note:** These are the **maximum** capabilities the server allows. The client must still report its own capabilities.
 
 ---
 
@@ -274,8 +313,60 @@ Used for DM pairing flow.
 
 ---
 
+#### 8. Capabilities Request
+
+```json
+{
+  "type": "request_capabilities"
+}
+```
+
+Sent by server **after handshake** to request client's capabilities.
+
+**Client must respond with `capabilities_response`** (see below).
+
+---
+
 ### Client → Server
 
+#### 0. Capabilities Response (required)
+
+**Sent in response to `request_capabilities` from server.**
+
+```json
+{
+  "type": "capabilities_response",
+  "capabilities": {
+    "text": true,
+    "images": true,
+    "files": false,
+    "voice": true,
+    "reactions": false,
+    "threads": true,
+    "typing": true
+  },
+  "metadata": {
+    "clientName": "Telegram Bot",
+    "version": "1.0.0",
+    "platform": "telegram"
+  }
+}
+```
+
+**Important:** Client **must** send this message after receiving `request_capabilities` from server.
+
+**Fields:**
+- `capabilities` - What **this client** supports (true = supported, false = not supported)
+- `metadata` - Optional client metadata (name, version, platform, etc.)
+
+**Effective Capabilities:**
+The server uses the **intersection** of server capabilities and client capabilities:
+- Server allows: `{ images: true, voice: false }`
+- Client supports: `{ images: true, voice: true }`
+- **Effective:** `{ images: true, voice: false }` (most restrictive wins)
+
+---
+
 #### 1. Text Message
 
 ```json
@@ -471,6 +562,10 @@ class OpenClawClient {
         this.handleMessage(message);
       });
 
+      this.ws.on('open', () => {
+        console.log('Connected to OpenClaw');
+      });
+
       this.ws.on('error', (error) => {
         reject(error);
       });
@@ -489,6 +584,27 @@ class OpenClawClient {
         console.log('Server capabilities:', message.capabilities);
         break;
 
+      case 'request_capabilities':
+        // Send client capabilities
+        this.ws.send(JSON.stringify({
+          type: 'capabilities_response',
+          capabilities: {
+            text: true,
+            images: this.supportsImages(),
+            files: this.supportsFiles(),
+            voice: this.supportsVoice(),
+            reactions: false,
+            threads: true,
+            typing: true
+          },
+          metadata: {
+            clientName: 'OpenClaw WebSocket Client',
+            version: '1.0.0'
+          }
+        }));
+        console.log('Sent client capabilities');
+        break;
+
       case 'message':
         // Bot sent a message
         this.messageHandlers.forEach(h => h(message));
@@ -509,6 +625,11 @@ class OpenClawClient {
     }
   }
 
+  // Override these methods based on your client's capabilities
+  supportsImages() { return true; }
+  supportsFiles() { return false; }
+  supportsVoice() { return false; }
+
   onMessage(handler) {
     this.messageHandlers.push(handler);
   }

src/channel.test.ts

@@ -188,18 +188,22 @@ describe("WebSocketServerManager", () => {
   });
 
   describe("addClient", () => {
-    it("adds a client and sends handshake", () => {
+    it("adds a client and sends handshake + capabilities request", () => {
       const sentMessages: string[] = [];
       mockWs.send = (data: string) => sentMessages.push(data);
 
       manager.addClient(mockWs as any, mockAccount);
 
-      expect(sentMessages.length).toBe(1);
+      expect(sentMessages.length).toBe(2); // handshake + request_capabilities
+      
       const handshake = JSON.parse(sentMessages[0]);
       expect(handshake.type).toBe("handshake");
       expect(handshake.accountId).toBe("test-account");
       expect(handshake.capabilities.text).toBe(true);
       expect(handshake.metadata.serverName).toBe("Test Server");
+
+      const capsRequest = JSON.parse(sentMessages[1]);
+      expect(capsRequest.type).toBe("request_capabilities");
     });
 
     it("tracks connection count", () => {
@@ -215,6 +219,121 @@ describe("WebSocketServerManager", () => {
     });
   });
 
+  describe("capabilities negotiation", () => {
+    it("receives and processes capabilities response", () => {
+      const sentMessages: string[] = [];
+      mockWs.send = (data: string) => sentMessages.push(data);
+
+      manager.addClient(mockWs as any, mockAccount);
+
+      // Simulate client capabilities response
+      mockWs.emit("message", Buffer.from(JSON.stringify({
+        type: "capabilities_response",
+        capabilities: {
+          text: true,
+          images: true,
+          files: false,
+          voice: false,
+          reactions: false,
+          threads: true,
+          typing: true
+        },
+        metadata: { clientName: "Test Client" }
+      })));
+
+      const connections = manager.getConnections();
+      expect(connections[0].capabilitiesReady).toBe(true);
+      expect(connections[0].clientCapabilities?.images).toBe(true);
+      expect(connections[0].clientCapabilities?.files).toBe(false);
+      expect(connections[0].clientMetadata?.clientName).toBe("Test Client");
+    });
+
+    it("uses intersection of account and client capabilities", () => {
+      const sentMessages: string[] = [];
+      mockWs.send = (data: string) => sentMessages.push(data);
+
+      // Account allows everything
+      const fullCapAccount = {
+        ...mockAccount,
+        capabilities: {
+          text: true,
+          images: true,
+          files: true,
+          voice: true,
+          reactions: true,
+          threads: true,
+          typing: true
+        }
+      };
+
+      manager.addClient(mockWs as any, fullCapAccount);
+
+      // Client only supports text and images
+      mockWs.emit("message", Buffer.from(JSON.stringify({
+        type: "capabilities_response",
+        capabilities: {
+          text: true,
+          images: true,
+          files: false,
+          voice: false,
+          reactions: false,
+          threads: false,
+          typing: false
+        }
+      })));
+
+      const effective = manager.getEffectiveCapabilities(manager.getConnections()[0]);
+      expect(effective.text).toBe(true);
+      expect(effective.images).toBe(true);
+      expect(effective.files).toBe(false); // Client doesn't support
+      expect(effective.voice).toBe(false); // Client doesn't support
+    });
+
+    it("falls back to account capabilities if client doesn't report", () => {
+      const sentMessages: string[] = [];
+      mockWs.send = (data: string) => sentMessages.push(data);
+
+      manager.addClient(mockWs as any, mockAccount);
+
+      // Don't send capabilities_response
+      const effective = manager.getEffectiveCapabilities(manager.getConnections()[0]);
+      
+      // Should use account capabilities
+      expect(effective.text).toBe(true);
+      expect(effective.images).toBe(false); // Account default
+    });
+
+    it("clientSupportsCapability checks effective capabilities", () => {
+      const sentMessages: string[] = [];
+      mockWs.send = (data: string) => sentMessages.push(data);
+
+      manager.addClient(mockWs as any, mockAccount);
+
+      // Before capabilities response - uses account defaults
+      expect(manager.clientSupportsCapability(manager.getConnections()[0], "text")).toBe(true);
+      expect(manager.clientSupportsCapability(manager.getConnections()[0], "images")).toBe(false);
+
+      // Send capabilities response
+      mockWs.emit("message", Buffer.from(JSON.stringify({
+        type: "capabilities_response",
+        capabilities: {
+          text: true,
+          images: true,
+          files: false,
+          voice: false,
+          reactions: false,
+          threads: false,
+          typing: false
+        }
+      })));
+
+      // After response - uses intersection
+      expect(manager.clientSupportsCapability(manager.getConnections()[0], "text")).toBe(true);
+      expect(manager.clientSupportsCapability(manager.getConnections()[0], "images")).toBe(true);
+      expect(manager.clientSupportsCapability(manager.getConnections()[0], "files")).toBe(false);
+    });
+  });
+
   describe("sendText", () => {
     it("sends text message to all connections for account", async () => {
       const sentMessages: string[] = [];

src/client.ts

@@ -61,18 +61,20 @@ async function handleInboundMessage(
     return;
   }
 
-  // Validate capabilities
+  // Validate capabilities - check both account and client capabilities
   const content = message.content;
-  if (content.type === "image" && !account.capabilities.images) {
+  
-    api.logger?.warn(`Account ${account.accountId} tried to send image but capability is disabled`);
+  // Check if CLIENT supports this capability (client-reported)
+  if (content.type === "image" && !wsManager.clientSupportsCapability(client, "images")) {
+    api.logger?.warn(`Client ${client.account.accountId} tried to send image but capability is not supported by client`);
     return;
   }
-  if (content.type === "file" && !account.capabilities.files) {
+  if (content.type === "file" && !wsManager.clientSupportsCapability(client, "files")) {
-    api.logger?.warn(`Account ${account.accountId} tried to send file but capability is disabled`);
+    api.logger?.warn(`Client ${client.account.accountId} tried to send file but capability is not supported by client`);
     return;
   }
-  if (content.type === "voice" && !account.capabilities.voice) {
+  if (content.type === "voice" && !wsManager.clientSupportsCapability(client, "voice")) {
-    api.logger?.warn(`Account ${account.accountId} tried to send voice but capability is disabled`);
+    api.logger?.warn(`Client ${client.account.accountId} tried to send voice but capability is not supported by client`);
     return;
   }
 

src/websocket-server.ts

@@ -45,12 +45,24 @@ export type PairingCodeMessage = {
   message: string;
 };
 
+export type CapabilitiesRequest = {
+  type: "request_capabilities";
+};
+
+export type CapabilitiesResponse = {
+  type: "capabilities_response";
+  capabilities: ChannelCapabilities;
+  metadata?: Record<string, unknown>;
+};
+
 export type WebSocketMessage =
   | InboundMessage
   | OutboundMessage
   | HandshakeMessage
   | TypingIndicator
-  | PairingCodeMessage;
+  | PairingCodeMessage
+  | CapabilitiesRequest
+  | CapabilitiesResponse;
 
 export type MessageContent =
   | { type: "text"; text: string }
@@ -65,6 +77,9 @@ export type ConnectedClient = {
   userId?: string;
   connectedAt: Date;
   lastPingAt: Date;
+  clientCapabilities?: ChannelCapabilities;  // Capabilities reported by client
+  clientMetadata?: Record<string, unknown>;  // Metadata reported by client
+  capabilitiesReady: boolean;  // Whether client has reported capabilities
 };
 
 export type MessageHandler = (message: InboundMessage, client: ConnectedClient) => Promise<void> | void;
@@ -89,15 +104,28 @@ export class WebSocketServerManager {
       account,
       connectedAt: new Date(),
       lastPingAt: new Date(),
+      clientCapabilities: undefined,
+      clientMetadata: undefined,
+      capabilitiesReady: false,
     };
     this.clients.set(ws, client);
 
     // Set up message handler
     ws.on("message", async (data) => {
       try {
-        const message = JSON.parse(data.toString()) as InboundMessage;
+        const message = JSON.parse(data.toString()) as WebSocketMessage;
-        if (this.messageHandler) {
+        
-          await this.messageHandler(message, client);
+        // Handle capabilities response from client
+        if (message.type === "capabilities_response") {
+          this.handleCapabilitiesResponse(client, message);
+          return;
+        }
+        
+        // Handle regular inbound messages
+        if (message.type === "message") {
+          if (this.messageHandler) {
+            await this.messageHandler(message as InboundMessage, client);
+          }
         }
       } catch (err) {
         console.error("Failed to parse WebSocket message:", err);
@@ -121,9 +149,52 @@ export class WebSocketServerManager {
     // Send handshake
     this.sendHandshake(client);
 
+    // Request client capabilities after handshake
+    this.requestCapabilities(client);
+
     return client;
   }
 
+  /**
+   * Request capabilities from client
+   */
+  private requestCapabilities(client: ConnectedClient): void {
+    const request: CapabilitiesRequest = {
+      type: "request_capabilities",
+    };
+    this.sendToClient(client.ws, request);
+  }
+
+  /**
+   * Handle capabilities response from client
+   */
+  private handleCapabilitiesResponse(
+    client: ConnectedClient,
+    response: CapabilitiesResponse
+  ): void {
+    // Merge client capabilities with account capabilities (intersection)
+    const accountCaps = client.account.capabilities;
+    const clientCaps = response.capabilities;
+
+    client.clientCapabilities = {
+      text: accountCaps.text && clientCaps.text,
+      images: accountCaps.images && clientCaps.images,
+      files: accountCaps.files && clientCaps.files,
+      voice: accountCaps.voice && clientCaps.voice,
+      reactions: accountCaps.reactions && clientCaps.reactions,
+      threads: accountCaps.threads && clientCaps.threads,
+      typing: accountCaps.typing && clientCaps.typing,
+    };
+
+    client.clientMetadata = response.metadata;
+    client.capabilitiesReady = true;
+
+    console.log(
+      `Client capabilities received for account ${client.account.accountId}:`,
+      client.clientCapabilities
+    );
+  }
+
   /**
    * Send handshake message to client
    */
@@ -131,13 +202,34 @@ export class WebSocketServerManager {
     const handshake: HandshakeMessage = {
       type: "handshake",
       accountId: client.account.accountId,
-      capabilities: client.account.capabilities,
+      capabilities: client.account.capabilities,  // Server capabilities (max allowed)
       metadata: client.account.metadata || {},
       timestamp: Date.now(),
     };
     this.sendToClient(client.ws, handshake);
   }
 
+  /**
+   * Get effective capabilities for a client (intersection of account and client capabilities)
+   */
+  getEffectiveCapabilities(client: ConnectedClient): ChannelCapabilities {
+    // If client hasn't reported capabilities yet, use account capabilities
+    if (!client.capabilitiesReady || !client.clientCapabilities) {
+      return client.account.capabilities;
+    }
+
+    // Return intersection (client can only use what both support)
+    return client.clientCapabilities;
+  }
+
+  /**
+   * Check if a client supports a specific capability
+   */
+  clientSupportsCapability(client: ConnectedClient, capability: keyof ChannelCapabilities): boolean {
+    const caps = this.getEffectiveCapabilities(client);
+    return caps[capability] === true;
+  }
+
   /**
    * Send a message to a specific client
    */