fix(ai): send model for async opencode prompts

apis/libai.lua

@@ -199,6 +199,15 @@ local function createAi(opts)
     return DEFAULT_LUA_EXEC_TIMEOUT_SECONDS;
   end
 
+  local function resolveModel(options)
+    local providerId = options.providerID or settingsLib.get('opencc.provider_id');
+    local modelId = options.modelID or settingsLib.get('opencc.model_id');
+    if isBlank(providerId) or isBlank(modelId) then
+      return nil, nil;
+    end
+    return providerId, modelId;
+  end
+
   local function resolveConfig(options)
     local url = options.serverUrl or settingsLib.get('opencc.server_url');
     if not url or url == '' then
@@ -206,16 +215,30 @@ local function createAi(opts)
     end
     local username = options.username or settingsLib.get('opencc.username') or 'opencode';
     local password = options.password or settingsLib.get('opencc.password') or '';
+    local providerId, modelId = resolveModel(options);
     return {
       url = trimTrailingSlash(url),
       username = username,
       password = password,
+      providerID = providerId,
+      modelID = modelId,
       timeoutSeconds = resolveTimeout(options),
       pollTimeoutSeconds = resolvePollTimeout(options),
       pollIntervalSeconds = resolvePollInterval(options),
     };
   end
 
+  local function buildPromptBody(cfg, messageId, prompt)
+    local body = {
+      messageID = messageId,
+      parts = { { type = 'text', text = prompt } },
+    };
+    if cfg.providerID and cfg.modelID then
+      body.model = { providerID = cfg.providerID, modelID = cfg.modelID };
+    end
+    return body;
+  end
+
   local function createMessageId()
     local t = math.floor(nowFunc() * 1000);
     return 'msg_' .. tostring(t) .. '_' .. tostring(math.random(100000, 999999));
@@ -411,10 +434,8 @@ local function createAi(opts)
     end
 
     local messageId = options.messageId or createMessageId();
-    local body, code = doPost(cfg, '/session/' .. sessionId .. '/prompt_async', {
-      messageID = messageId,
-      parts = { { type = 'text', text = prompt } },
-    });
+    local body, code = doPost(cfg, '/session/' .. sessionId .. '/prompt_async',
+      buildPromptBody(cfg, messageId, prompt));
     if not body then return false, code; end
     if code == 404 then
       return handleMissingSession(persist, sessionSettingKey);
@@ -520,6 +541,8 @@ local function createAi(opts)
         serverUrl = options.serverUrl,
         username = options.username,
         password = options.password,
+        providerID = options.providerID,
+        modelID = options.modelID,
         timeoutSeconds = options.timeoutSeconds,
       };
     end

docs/opencode_api.md

@@ -132,6 +132,19 @@ Abort a running generation.
 
 Fire-and-forget variant. Returns `204` immediately. Include `messageID` in the request body to make the assistant response addressable by `GET /session/:id/message/:messageID`. Opencode validates caller-provided message IDs; use IDs starting with `msg`.
 
+**Request body:**
+```json
+{
+  "messageID": "msg_xyz",
+  "parts": [
+    { "type": "text", "text": "your prompt here" }
+  ],
+  "model": { "providerID": "anthropic", "modelID": "claude-opus-4-7" }
+}
+```
+
+Unlike `/message`, `model` is **not** optional in practice — omitting it causes the request to be accepted (`204`) without triggering generation, so the assistant message never appears. Always send `providerID` / `modelID`.
+
 `ai` uses this endpoint by default to avoid `504` failures from the blocking `/message` endpoint when the LLM takes longer than one HTTP request timeout. The submitted `messageID` identifies the user message; `ai` polls `GET /session/:id/message` and reads the completed assistant message that follows it.
 
 ---

docs/opencode_server_guide.md

@@ -90,6 +90,13 @@ Optional — override the Basic Auth username (default `opencode`):
 set opencc.username myuser
 ```
 
+Pick the provider and model. `ai` posts to `/session/:id/prompt_async`, which (unlike the blocking `/message` endpoint) does **not** fall back to a server-side default — the assistant message will never be generated if these are unset:
+
+```sh
+set opencc.provider_id anthropic
+set opencc.model_id claude-opus-4-7
+```
+
 - **CraftOS-PC (localhost):** `http://127.0.0.1:4096`
 - **In-game ATM10:** use your LAN IP (e.g. `192.168.x.x`) — add it to `http.rules` in `config/computercraft-server.toml`
 
@@ -124,5 +131,5 @@ Set settings inside the harness before running, or inject them via the test API.
 | `missing prompt` | No prompt was passed | Run `ai <prompt>` or `ai ping` |
 | `session introuvable; lance: ai new <prompt>` | Session was deleted or server restarted | Run `ai new <prompt>` |
 | `erreur message: HTTP 504` | AI took too long | Retry; consider a faster model |
-| `delai depasse en attendant la reponse AI` | Async polling timed out | Increase `opencc.poll_timeout_seconds` or check opencode logs |
+| `delai depasse en attendant la reponse AI` | Async polling timed out — often because opencode received the prompt but never started generation | Make sure `opencc.provider_id` and `opencc.model_id` are set; otherwise increase `opencc.poll_timeout_seconds` or check opencode logs |
 | `reponse vide` | Reply had no text parts | Check opencode logs |

manifest.json

@@ -1,6 +1,6 @@
 {
   "name": "TrapOS",
-  "version": "0.8.0",
+  "version": "0.8.1",
   "branch": "next",
   "packages": [
     "trapos"

packages/index.json

@@ -5,8 +5,8 @@
     "trapos-boot": "0.3.0",
     "trapos-net": "0.3.0",
     "trapos-ui": "0.2.2",
-    "trapos-ai": "0.6.0",
+    "trapos-ai": "0.6.1",
     "trapos-sandbox": "0.1.0",
-    "trapos": "0.8.0"
+    "trapos": "0.8.1"
   }
 }

packages/trapos-ai/ccpm.json

@@ -1,6 +1,6 @@
 {
   "name": "trapos-ai",
-  "version": "0.6.0",
+  "version": "0.6.1",
   "description": "TrapOS AI client for opencode serve",
   "dependencies": ["trapos-core"],
   "files": [

packages/trapos/ccpm.json

@@ -1,6 +1,6 @@
 {
   "name": "trapos",
-  "version": "0.8.0",
+  "version": "0.8.1",
   "description": "TrapOS full install meta-package",
   "dependencies": ["trapos-boot", "trapos-net", "trapos-ui", "trapos-test", "trapos-ai"],
   "files": [],

programs/ai.lua

@@ -24,6 +24,8 @@ local function printUsage()
   print('  opencc.username    (default: opencode)');
   print('  opencc.password    (Basic Auth password)');
   print('  opencc.session_id  (auto-managed)');
+  print('  opencc.provider_id (e.g. anthropic)');
+  print('  opencc.model_id    (e.g. claude-opus-4-7)');
   print('  opencc.timeout_seconds       (per HTTP call, max 60)');
   print('  opencc.poll_timeout_seconds  (default: 300)');
   print('  opencc.poll_interval_seconds (default: 2)');

tests/ai.lua

@@ -346,6 +346,99 @@ testlib.test('ask sends exact prompt text', function()
   testlib.assertEquals(body.parts[1].text, 'my prompt');
 end);
 
+testlib.test('ask includes model when provider_id and model_id are set', function()
+  local httpStub = fakeHttp(
+    { messageResp('reply') },
+    {}
+  );
+  local settingsStub = fakeSettings({
+    ['opencc.server_url'] = 'http://host',
+    ['opencc.session_id'] = 'ses_1',
+    ['opencc.provider_id'] = 'anthropic',
+    ['opencc.model_id'] = 'claude-opus-4-7',
+  });
+  local ai = createAi({ http = httpStub, settings = settingsStub });
+
+  ai.ask('hello');
+
+  local body = textutils.unserializeJSON(httpStub.postCalls[1].body);
+  testlib.assertEquals(body.model.providerID, 'anthropic');
+  testlib.assertEquals(body.model.modelID, 'claude-opus-4-7');
+end);
+
+testlib.test('ask omits model when provider_id is missing', function()
+  local httpStub = fakeHttp(
+    { messageResp('reply') },
+    {}
+  );
+  local settingsStub = fakeSettings({
+    ['opencc.server_url'] = 'http://host',
+    ['opencc.session_id'] = 'ses_1',
+    ['opencc.model_id'] = 'claude-opus-4-7',
+  });
+  local ai = createAi({ http = httpStub, settings = settingsStub });
+
+  ai.ask('hello');
+
+  local body = textutils.unserializeJSON(httpStub.postCalls[1].body);
+  testlib.assertEquals(body.model, nil);
+end);
+
+testlib.test('ask omits model when model_id is missing', function()
+  local httpStub = fakeHttp(
+    { messageResp('reply') },
+    {}
+  );
+  local settingsStub = fakeSettings({
+    ['opencc.server_url'] = 'http://host',
+    ['opencc.session_id'] = 'ses_1',
+    ['opencc.provider_id'] = 'anthropic',
+  });
+  local ai = createAi({ http = httpStub, settings = settingsStub });
+
+  ai.ask('hello');
+
+  local body = textutils.unserializeJSON(httpStub.postCalls[1].body);
+  testlib.assertEquals(body.model, nil);
+end);
+
+testlib.test('ask omits model when neither provider_id nor model_id is set', function()
+  local httpStub = fakeHttp(
+    { messageResp('reply') },
+    {}
+  );
+  local settingsStub = fakeSettings({
+    ['opencc.server_url'] = 'http://host',
+    ['opencc.session_id'] = 'ses_1',
+  });
+  local ai = createAi({ http = httpStub, settings = settingsStub });
+
+  ai.ask('hello');
+
+  local body = textutils.unserializeJSON(httpStub.postCalls[1].body);
+  testlib.assertEquals(body.model, nil);
+end);
+
+testlib.test('ask options providerID/modelID override settings', function()
+  local httpStub = fakeHttp(
+    { messageResp('reply') },
+    {}
+  );
+  local settingsStub = fakeSettings({
+    ['opencc.server_url'] = 'http://host',
+    ['opencc.session_id'] = 'ses_1',
+    ['opencc.provider_id'] = 'anthropic',
+    ['opencc.model_id'] = 'claude-opus-4-7',
+  });
+  local ai = createAi({ http = httpStub, settings = settingsStub });
+
+  ai.ask('hello', { providerID = 'openai', modelID = 'gpt-5' });
+
+  local body = textutils.unserializeJSON(httpStub.postCalls[1].body);
+  testlib.assertEquals(body.model.providerID, 'openai');
+  testlib.assertEquals(body.model.modelID, 'gpt-5');
+end);
+
 testlib.test('ask generates opencode-compatible message ids', function()
   local httpStub = fakeHttp(
     { messageResp('reply') },