feat(rlsapi): include system info context in LLM instructions

Enhance the LLM instructions with the user's RHEL system information
(OS, version, architecture) when available. This gives the LLM better
context about the environment the user is asking questions about,
enabling more relevant and accurate responses.

Signed-off-by: Major Hayden <major@redhat.com>

Author: major

src/app/endpoints/rlsapi_v1.py

@@ -27,7 +27,7 @@
     UnauthorizedResponse,
     UnprocessableEntityResponse,
 )
-from models.rlsapi.requests import RlsapiV1InferRequest
+from models.rlsapi.requests import RlsapiV1InferRequest, RlsapiV1SystemInfo
 from models.rlsapi.responses import RlsapiV1InferData, RlsapiV1InferResponse
 from utils.responses import extract_text_from_response_output_item
 from utils.suid import get_suid
@@ -49,6 +49,35 @@
 }
 
 
+def _build_instructions(systeminfo: RlsapiV1SystemInfo) -> str:
+    """Build LLM instructions incorporating system context when available.
+
+    Enhances the default system prompt with RHEL system information to provide
+    the LLM with relevant context about the user's environment.
+
+    Args:
+        systeminfo: System information from the client (OS, version, arch).
+
+    Returns:
+        Instructions string for the LLM, with system context if available.
+    """
+    base_prompt = constants.DEFAULT_SYSTEM_PROMPT
+
+    context_parts = []
+    if systeminfo.os:
+        context_parts.append(f"OS: {systeminfo.os}")
+    if systeminfo.version:
+        context_parts.append(f"Version: {systeminfo.version}")
+    if systeminfo.arch:
+        context_parts.append(f"Architecture: {systeminfo.arch}")
+
+    if not context_parts:
+        return base_prompt
+
+    system_context = ", ".join(context_parts)
+    return f"{base_prompt}\n\nUser's system: {system_context}"
+
+
 def _get_default_model_id() -> str:
     """Get the default model ID from configuration.
 
@@ -82,14 +111,15 @@ def _get_default_model_id() -> str:
     )
 
 
-async def retrieve_simple_response(question: str) -> str:
+async def retrieve_simple_response(question: str, instructions: str) -> str:
     """Retrieve a simple response from the LLM for a stateless query.
 
     Uses the Responses API for simple stateless inference, consistent with
     other endpoints (query_v2, streaming_query_v2).
 
     Args:
         question: The combined user input (question + context).
+        instructions: System instructions for the LLM.
 
     Returns:
         The LLM-generated response text.
@@ -106,7 +136,7 @@ async def retrieve_simple_response(question: str) -> str:
     response = await client.responses.create(
         input=question,
         model=model_id,
-        instructions=constants.DEFAULT_SYSTEM_PROMPT,
+        instructions=instructions,
         stream=False,
         store=False,
     )
@@ -149,14 +179,14 @@ async def infer_endpoint(
 
     logger.info("Processing rlsapi v1 /infer request %s", request_id)
 
-    # Combine all input sources (question, stdin, attachments, terminal)
     input_source = infer_request.get_input_source()
+    instructions = _build_instructions(infer_request.context.systeminfo)
     logger.debug(
         "Request %s: Combined input source length: %d", request_id, len(input_source)
     )
 
     try:
-        response_text = await retrieve_simple_response(input_source)
+        response_text = await retrieve_simple_response(input_source, instructions)
     except APIConnectionError as e:
         metrics.llm_calls_failures_total.inc()
         logger.error(

tests/unit/app/endpoints/test_rlsapi_v1.py

@@ -13,6 +13,7 @@
 
 import constants
 from app.endpoints.rlsapi_v1 import (
+    _build_instructions,
     _get_default_model_id,
     infer_endpoint,
     retrieve_simple_response,
@@ -96,6 +97,35 @@ def mock_api_connection_error_fixture(mocker: MockerFixture) -> None:
     )
 
 
+# --- Test _build_instructions ---
+
+
+def test_build_instructions_with_full_systeminfo() -> None:
+    """Test _build_instructions includes all system info fields."""
+    systeminfo = RlsapiV1SystemInfo(os="RHEL", version="9.3", arch="x86_64")
+    result = _build_instructions(systeminfo)
+    assert constants.DEFAULT_SYSTEM_PROMPT in result
+    assert "OS: RHEL" in result
+    assert "Version: 9.3" in result
+    assert "Architecture: x86_64" in result
+
+
+def test_build_instructions_with_partial_systeminfo() -> None:
+    """Test _build_instructions handles partial system info."""
+    systeminfo = RlsapiV1SystemInfo(os="RHEL", version="", arch="")
+    result = _build_instructions(systeminfo)
+    assert "OS: RHEL" in result
+    assert "Version:" not in result
+    assert "Architecture:" not in result
+
+
+def test_build_instructions_with_empty_systeminfo() -> None:
+    """Test _build_instructions returns base prompt when no system info."""
+    systeminfo = RlsapiV1SystemInfo()
+    result = _build_instructions(systeminfo)
+    assert result == constants.DEFAULT_SYSTEM_PROMPT
+
+
 # --- Test _get_default_model_id ---
 
 
@@ -151,7 +181,9 @@ async def test_retrieve_simple_response_success(
     mock_configuration: AppConfig, mock_llm_response: None
 ) -> None:
     """Test retrieve_simple_response returns LLM response text."""
-    response = await retrieve_simple_response("How do I list files?")
+    response = await retrieve_simple_response(
+        "How do I list files?", constants.DEFAULT_SYSTEM_PROMPT
+    )
     assert response == "This is a test LLM response."
 
 
@@ -160,7 +192,9 @@ async def test_retrieve_simple_response_empty_output(
     mock_configuration: AppConfig, mock_empty_llm_response: None
 ) -> None:
     """Test retrieve_simple_response handles empty LLM output."""
-    response = await retrieve_simple_response("Test question")
+    response = await retrieve_simple_response(
+        "Test question", constants.DEFAULT_SYSTEM_PROMPT
+    )
     assert response == ""
 
 
@@ -170,7 +204,7 @@ async def test_retrieve_simple_response_api_connection_error(
 ) -> None:
     """Test retrieve_simple_response propagates APIConnectionError."""
     with pytest.raises(APIConnectionError):
-        await retrieve_simple_response("Test question")
+        await retrieve_simple_response("Test question", constants.DEFAULT_SYSTEM_PROMPT)
 
 
 # --- Test infer_endpoint ---