fix(api): forward real LLM usage in /v1/chat/completions response (#4310)

* fix(api): forward real LLM usage in /v1/chat/completions response

_chat_completion_response() hardcoded prompt_tokens/completion_tokens
to zero.  Now reads agent_loop._last_usage (set by process_direct
after every LLM call) and forwards the actual prompt/completion counts.

Streaming path is unchanged; usage is only surfaced in non-streaming
responses for now.

Fixes #4309

* fix: use defensive getattr for _last_usage and add it to all test mock agents

- Use getattr(agent_loop, '_last_usage', None) in server.py for safety
- Add _last_usage = {} to mock agents in test_api_attachment.py and test_api_stream.py
- Prevents AttributeError/500 when mock agents don't have the attribute

* fix(api): preserve provider total usage

---------

Co-authored-by: michaelxer <michaelxer@users.noreply.github.com>
Co-authored-by: Xubin Ren <52506698+Re-bin@users.noreply.github.com>

nanobot/api/server.py

@@ -54,7 +54,14 @@ def _error_json(status: int, message: str, err_type: str = "invalid_request_erro
     )
 
 
-def _chat_completion_response(content: str, model: str) -> dict[str, Any]:
+def _chat_completion_response(
+    content: str,
+    model: str,
+    usage: dict[str, int] | None = None,
+) -> dict[str, Any]:
+    prompt = (usage or {}).get("prompt_tokens", 0)
+    completion = (usage or {}).get("completion_tokens", 0)
+    total = (usage or {}).get("total_tokens", 0) or prompt + completion
     return {
         "id": f"chatcmpl-{uuid.uuid4().hex[:12]}",
         "object": "chat.completion",
@@ -67,7 +74,11 @@ def _chat_completion_response(content: str, model: str) -> dict[str, Any]:
                 "finish_reason": "stop",
             }
         ],
-        "usage": {"prompt_tokens": 0, "completion_tokens": 0, "total_tokens": 0},
+        "usage": {
+            "prompt_tokens": prompt,
+            "completion_tokens": completion,
+            "total_tokens": total,
+        },
     }
 
 
@@ -346,7 +357,9 @@ async def handle_chat_completions(request: web.Request) -> web.Response:
         logger.exception("Unexpected API lock error for session {}", session_key)
         return _error_json(500, "Internal server error", err_type="server_error")
 
-    return web.json_response(_chat_completion_response(response_text, model_name))
+    return web.json_response(
+        _chat_completion_response(response_text, model_name, getattr(agent_loop, "_last_usage", None))
+    )
 
 
 async def handle_models(request: web.Request) -> web.Response:

tests/test_api_attachment.py

@@ -32,6 +32,7 @@ def _make_mock_agent(response_text: str = "mock response") -> MagicMock:
     agent.process_direct = AsyncMock(return_value=response_text)
     agent._connect_mcp = AsyncMock()
     agent.close_mcp = AsyncMock()
+    agent._last_usage = {}
     return agent
 
 

tests/test_api_stream.py

@@ -75,6 +75,7 @@ def _make_streaming_agent(tokens: list[str]) -> MagicMock:
         return " ".join(tokens)
 
     agent.process_direct = fake_process_direct
+    agent._last_usage = {}
     return agent
 
 
@@ -133,6 +134,7 @@ async def test_stream_false_returns_json(aiohttp_client) -> None:
     agent.process_direct = AsyncMock(return_value="normal reply")
     agent._connect_mcp = AsyncMock()
     agent.close_mcp = AsyncMock()
+    agent._last_usage = {}
 
     app = create_app(agent, model_name="m")
     client = await aiohttp_client(app)
@@ -155,6 +157,7 @@ async def test_stream_default_is_false(aiohttp_client) -> None:
     agent.process_direct = AsyncMock(return_value="default reply")
     agent._connect_mcp = AsyncMock()
     agent.close_mcp = AsyncMock()
+    agent._last_usage = {}
 
     app = create_app(agent, model_name="m")
     client = await aiohttp_client(app)
@@ -209,6 +212,7 @@ async def test_stream_passes_on_stream_callbacks(aiohttp_client) -> None:
     agent.process_direct = fake_process_direct
     agent._connect_mcp = AsyncMock()
     agent.close_mcp = AsyncMock()
+    agent._last_usage = {}
 
     app = create_app(agent, model_name="m")
     client = await aiohttp_client(app)
@@ -241,6 +245,7 @@ async def test_stream_segment_end_does_not_close_sse(aiohttp_client) -> None:
     agent.process_direct = fake_process_direct
     agent._connect_mcp = AsyncMock()
     agent.close_mcp = AsyncMock()
+    agent._last_usage = {}
 
     app = create_app(agent, model_name="m")
     client = await aiohttp_client(app)
@@ -279,6 +284,7 @@ async def test_stream_uses_final_response_when_no_deltas(aiohttp_client) -> None
     agent.process_direct = fake_process_direct
     agent._connect_mcp = AsyncMock()
     agent.close_mcp = AsyncMock()
+    agent._last_usage = {}
 
     app = create_app(agent, model_name="m")
     client = await aiohttp_client(app)
@@ -320,6 +326,7 @@ async def test_stream_with_session_id(aiohttp_client) -> None:
     agent.process_direct = fake_process_direct
     agent._connect_mcp = AsyncMock()
     agent.close_mcp = AsyncMock()
+    agent._last_usage = {}
 
     app = create_app(agent, model_name="m")
     client = await aiohttp_client(app)
@@ -348,6 +355,7 @@ async def test_streaming_backend_failure_does_not_emit_success_terminator(aiohtt
     agent.process_direct = boom
     agent._connect_mcp = AsyncMock()
     agent.close_mcp = AsyncMock()
+    agent._last_usage = {}
 
     app = create_app(agent, model_name="m")
     client = await aiohttp_client(app)

tests/test_openai_api.py

@@ -33,6 +33,7 @@ def _make_mock_agent(response_text: str = "mock response") -> MagicMock:
     agent.process_direct = AsyncMock(return_value=response_text)
     agent._connect_mcp = AsyncMock()
     agent.close_mcp = AsyncMock()
+    agent._last_usage = {"prompt_tokens": 100, "completion_tokens": 50}
     return agent
 
 
@@ -78,6 +79,25 @@ def test_chat_completion_response() -> None:
     assert result["choices"][0]["message"]["content"] == "hello world"
     assert result["choices"][0]["finish_reason"] == "stop"
     assert result["id"].startswith("chatcmpl-")
+    assert result["usage"]["prompt_tokens"] == 0
+    assert result["usage"]["completion_tokens"] == 0
+    assert result["usage"]["total_tokens"] == 0
+
+
+def test_chat_completion_response_with_usage() -> None:
+    usage = {"prompt_tokens": 150, "completion_tokens": 42}
+    result = _chat_completion_response("hello world", "test-model", usage)
+    assert result["usage"]["prompt_tokens"] == 150
+    assert result["usage"]["completion_tokens"] == 42
+    assert result["usage"]["total_tokens"] == 192
+
+
+def test_chat_completion_response_preserves_provider_total_usage() -> None:
+    usage = {"total_tokens": 77}
+    result = _chat_completion_response("hello world", "test-model", usage)
+    assert result["usage"]["prompt_tokens"] == 0
+    assert result["usage"]["completion_tokens"] == 0
+    assert result["usage"]["total_tokens"] == 77
 
 
 @pytest.mark.skipif(not HAS_AIOHTTP, reason="aiohttp not installed")
@@ -213,6 +233,7 @@ async def test_followup_requests_share_same_session_key(aiohttp_client) -> None:
     agent.process_direct = fake_process
     agent._connect_mcp = AsyncMock()
     agent.close_mcp = AsyncMock()
+    agent._last_usage = {}
 
     app = create_app(agent, model_name="m")
     client = await aiohttp_client(app)
@@ -250,6 +271,7 @@ async def test_fixed_session_requests_are_serialized(aiohttp_client) -> None:
     agent.process_direct = slow_process
     agent._connect_mcp = AsyncMock()
     agent.close_mcp = AsyncMock()
+    agent._last_usage = {}
 
     app = create_app(agent, model_name="m")
     client = await aiohttp_client(app)
@@ -364,6 +386,7 @@ async def test_empty_response_retry_then_success(aiohttp_client) -> None:
     agent.process_direct = sometimes_empty
     agent._connect_mcp = AsyncMock()
     agent.close_mcp = AsyncMock()
+    agent._last_usage = {}
 
     app = create_app(agent, model_name="m")
     client = await aiohttp_client(app)
@@ -393,6 +416,7 @@ async def test_empty_response_falls_back(aiohttp_client) -> None:
     agent.process_direct = always_empty
     agent._connect_mcp = AsyncMock()
     agent.close_mcp = AsyncMock()
+    agent._last_usage = {}
 
     app = create_app(agent, model_name="m")
     client = await aiohttp_client(app)