fix(daemon): preserve runtime usage attribution in diagnostics (#7076)

Author: lefarcen

apps/daemon/src/runtimes/runs.ts

@@ -203,6 +203,8 @@ function summarizeAssistantMessageEvents(events) {
   let upstreamErrorCount = 0;
   let provider;
   let model;
+  let usageProvider;
+  let usageModel;
   let fallbackOrdinal = 0;
   const countErrorClass = (value) => {
     if (value === 'rate_limited') rateLimitedCount += 1;
@@ -212,6 +214,15 @@ function summarizeAssistantMessageEvents(events) {
   for (const record of events) {
     if (record?.event !== 'agent' || !record.data || typeof record.data !== 'object') continue;
     const data = record.data;
+    if (data.type === 'usage') {
+      if (typeof data.provider === 'string' && data.provider.trim()) {
+        usageProvider = data.provider.trim();
+      }
+      if (typeof data.model === 'string' && data.model.trim()) {
+        usageModel = data.model.trim();
+      }
+      continue;
+    }
     if (data.type !== 'diagnostic') continue;
     if (data.name === 'model_retry') {
       retryCount += 1;
@@ -282,8 +293,8 @@ function summarizeAssistantMessageEvents(events) {
     rateLimitedCount,
     timeoutCount,
     upstreamErrorCount,
-    provider,
-    model,
+    provider: provider ?? usageProvider,
+    model: model ?? usageModel,
   };
 }
 

apps/daemon/tests/runtimes/runs.test.ts

@@ -155,6 +155,64 @@ describe('chat run service shutdown', () => {
     vi.useRealTimers();
   });
 
+  it('uses runtime usage attribution when the adapter has no message lifecycle', () => {
+    const runs = createRuns();
+    const run = runs.create({
+      projectId: 'project-1',
+      conversationId: 'conv-1',
+      agentId: 'deepseek-harness',
+    }) as any;
+    run.model = 'default';
+    runs.emit(run, 'agent', {
+      type: 'usage',
+      provider: 'deepseek-official',
+      model: 'deepseek-v4-flash',
+      usage: { input_tokens: 10, output_tokens: 2 },
+    });
+    runs.finish(run, 'succeeded', 0, null);
+
+    const diagnostics = runs.statusBody(run).executionDiagnostics;
+    expect(diagnostics?.environment).toMatchObject({
+      requestedModel: { state: 'available', value: 'default' },
+      provider: { state: 'available', value: 'deepseek-official' },
+      resolvedModel: { state: 'available', value: 'deepseek-v4-flash' },
+    });
+    expect(diagnostics?.assistantMessages.count).toMatchObject({
+      state: 'not_collected',
+      missingReason: 'assistant_message_lifecycle_not_exposed_by_runtime',
+    });
+  });
+
+  it('keeps lifecycle attribution ahead of the usage fallback', () => {
+    const runs = createRuns();
+    const run = runs.create({
+      projectId: 'project-1',
+      conversationId: 'conv-1',
+      agentId: 'amr',
+    }) as any;
+    runs.emit(run, 'agent', {
+      type: 'usage',
+      provider: 'usage-provider',
+      model: 'usage-model',
+      usage: { input_tokens: 10, output_tokens: 2 },
+    });
+    runs.emit(run, 'agent', {
+      type: 'diagnostic',
+      name: 'assistant_message_lifecycle',
+      phase: 'end',
+      status: 'completed',
+      assistantMessageIndex: 1,
+      provider: 'lifecycle-provider',
+      model: 'lifecycle-model',
+    });
+    runs.finish(run, 'succeeded', 0, null);
+
+    expect(runs.statusBody(run).executionDiagnostics?.environment).toMatchObject({
+      provider: { state: 'available', value: 'lifecycle-provider' },
+      resolvedModel: { state: 'available', value: 'lifecycle-model' },
+    });
+  });
+
   it('keeps model-step percentiles unavailable until the documented sample minimum', () => {
     const runs = createRuns();
     const run = runs.create({ projectId: 'project-1', conversationId: 'conv-1', agentId: 'amr' }) as any;