diff --git a/AGENTS.md b/AGENTS.md
index 8ab8793c..89213e7d 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -37,6 +37,22 @@ Run the Flask API (if needed):
 flask --app application/app.py run --host=0.0.0.0 --port=7091
 ```
 
+That's the fast inner-loop option — quick startup, the Werkzeug interactive
+debugger still works, and it hot-reloads on source changes. It serves the
+Flask routes only (`/api/*`, `/stream`, etc.).
+
+If you need to exercise the full ASGI stack — the `/mcp` FastMCP endpoint,
+or to match the production runtime exactly — run the ASGI composition under
+uvicorn instead:
+
+```bash
+uvicorn application.asgi:asgi_app --host 0.0.0.0 --port 7091 --reload
+```
+
+Production uses `gunicorn -k uvicorn_worker.UvicornWorker` against the same
+`application.asgi:asgi_app` target; see `application/Dockerfile` for the
+full flag set.
+
 Run the Celery worker in a separate terminal (if needed):
 
 ```bash
diff --git a/application/Dockerfile b/application/Dockerfile
index 48d29e57..849bcc79 100644
--- a/application/Dockerfile
+++ b/application/Dockerfile
@@ -88,5 +88,15 @@ EXPOSE 7091
 # Switch to non-root user
 USER appuser
 
-# Start Gunicorn
-CMD ["gunicorn", "-w", "1", "--timeout", "120", "--bind", "0.0.0.0:7091", "--preload", "application.wsgi:app"]
+CMD ["gunicorn", \
+     "-w", "1", \
+     "-k", "uvicorn_worker.UvicornWorker", \
+     "--bind", "0.0.0.0:7091", \
+     "--timeout", "180", \
+     "--graceful-timeout", "120", \
+     "--keep-alive", "5", \
+     "--worker-tmp-dir", "/dev/shm", \
+     "--max-requests", "1000", \
+     "--max-requests-jitter", "100", \
+     "--config", "application/gunicorn_conf.py", \
+     "application.asgi:asgi_app"]
diff --git a/application/api/answer/routes/search.py b/application/api/answer/routes/search.py
index c5cdfaf8..d7aa2377 100644
--- a/application/api/answer/routes/search.py
+++ b/application/api/answer/routes/search.py
@@ -1,21 +1,21 @@
 import logging
-from typing import Any, Dict, List
 
 from flask import make_response, request
 from flask_restx import fields, Resource
 
 from application.api.answer.routes.base import answer_ns
-from application.core.settings import settings
-from application.storage.db.repositories.agents import AgentsRepository
-from application.storage.db.session import db_readonly
-from application.vectorstore.vector_creator import VectorCreator
+from application.services.search_service import (
+    InvalidAPIKey,
+    SearchFailed,
+    search,
+)
 
 logger = logging.getLogger(__name__)
 
 
 @answer_ns.route("/api/search")
 class SearchResource(Resource):
-    """Fast search endpoint for retrieving relevant documents"""
+    """Fast search endpoint for retrieving relevant documents."""
 
     search_model = answer_ns.model(
         "SearchModel",
@@ -32,102 +32,10 @@ class SearchResource(Resource):
         },
     )
 
-    def _get_sources_from_api_key(self, api_key: str) -> List[str]:
-        """Get source IDs connected to the API key/agent."""
-        with db_readonly() as conn:
-            agent_data = AgentsRepository(conn).find_by_key(api_key)
-        if not agent_data:
-            return []
-
-        source_ids: List[str] = []
-        # extra_source_ids is a PG ARRAY(UUID) of source UUIDs.
-        extra = agent_data.get("extra_source_ids") or []
-        for src in extra:
-            if src:
-                source_ids.append(str(src))
-
-        if not source_ids:
-            single = agent_data.get("source_id")
-            if single:
-                source_ids.append(str(single))
-
-        return source_ids
-
-    def _search_vectorstores(
-        self, query: str, source_ids: List[str], chunks: int
-    ) -> List[Dict[str, Any]]:
-        """Search across vectorstores and return results"""
-        if not source_ids:
-            return []
-
-        results = []
-        chunks_per_source = max(1, chunks // len(source_ids))
-        seen_texts = set()
-
-        for source_id in source_ids:
-            if not source_id or not source_id.strip():
-                continue
-
-            try:
-                docsearch = VectorCreator.create_vectorstore(
-                    settings.VECTOR_STORE, source_id, settings.EMBEDDINGS_KEY
-                )
-                docs = docsearch.search(query, k=chunks_per_source * 2)
-
-                for doc in docs:
-                    if len(results) >= chunks:
-                        break
-
-                    if hasattr(doc, "page_content") and hasattr(doc, "metadata"):
-                        page_content = doc.page_content
-                        metadata = doc.metadata
-                    else:
-                        page_content = doc.get("text", doc.get("page_content", ""))
-                        metadata = doc.get("metadata", {})
-
-                    # Skip duplicates
-                    text_hash = hash(page_content[:200])
-                    if text_hash in seen_texts:
-                        continue
-                    seen_texts.add(text_hash)
-
-                    title = metadata.get(
-                        "title", metadata.get("post_title", "")
-                    )
-                    if not isinstance(title, str):
-                        title = str(title) if title else ""
-
-                    # Clean up title
-                    if title:
-                        title = title.split("/")[-1]
-                    else:
-                        # Use filename or first part of content as title
-                        title = metadata.get("filename", page_content[:50] + "...")
-
-                    source = metadata.get("source", source_id)
-
-                    results.append({
-                        "text": page_content,
-                        "title": title,
-                        "source": source,
-                    })
-
-                if len(results) >= chunks:
-                    break
-
-            except Exception as e:
-                logger.error(
-                    f"Error searching vectorstore {source_id}: {e}",
-                    exc_info=True,
-                )
-                continue
-
-        return results[:chunks]
-
     @answer_ns.expect(search_model)
     @answer_ns.doc(description="Search for relevant documents based on query")
     def post(self):
-        data = request.get_json()
+        data = request.get_json() or {}
 
         question = data.get("question")
         api_key = data.get("api_key")
@@ -135,32 +43,13 @@ class SearchResource(Resource):
 
         if not question:
             return make_response({"error": "question is required"}, 400)
-
         if not api_key:
             return make_response({"error": "api_key is required"}, 400)
 
-        # Validate API key
-        with db_readonly() as conn:
-            agent = AgentsRepository(conn).find_by_key(api_key)
-        if not agent:
-            return make_response({"error": "Invalid API key"}, 401)
-
         try:
-            # Get sources connected to this API key
-            source_ids = self._get_sources_from_api_key(api_key)
-
-            if not source_ids:
-                return make_response([], 200)
-
-            # Perform search
-            results = self._search_vectorstores(question, source_ids, chunks)
-
-            return make_response(results, 200)
-
-        except Exception as e:
-            logger.error(
-                f"/api/search - error: {str(e)}",
-                extra={"error": str(e)},
-                exc_info=True,
-            )
+            return make_response(search(api_key, question, chunks), 200)
+        except InvalidAPIKey:
+            return make_response({"error": "Invalid API key"}, 401)
+        except SearchFailed:
+            logger.exception("/api/search failed")
             return make_response({"error": "Search failed"}, 500)
diff --git a/application/app.py b/application/app.py
index adc55ba5..31aef718 100644
--- a/application/app.py
+++ b/application/app.py
@@ -4,7 +4,7 @@ import platform
 import uuid
 
 import dotenv
-from flask import Flask, jsonify, redirect, request
+from flask import Flask, Response, jsonify, redirect, request
 from jose import jwt
 
 from application.auth import handle_auth
@@ -149,12 +149,11 @@ def authenticate_request():
 
 
 @app.after_request
-def after_request(response):
-    response.headers.add("Access-Control-Allow-Origin", "*")
-    response.headers.add("Access-Control-Allow-Headers", "Content-Type, Authorization")
-    response.headers.add(
-        "Access-Control-Allow-Methods", "GET, POST, PUT, DELETE, OPTIONS"
-    )
+def after_request(response: Response) -> Response:
+    """Add CORS headers for the pure Flask development entrypoint."""
+    response.headers["Access-Control-Allow-Origin"] = "*"
+    response.headers["Access-Control-Allow-Headers"] = "Content-Type, Authorization"
+    response.headers["Access-Control-Allow-Methods"] = "GET, POST, PUT, DELETE, OPTIONS"
     return response
 
 
diff --git a/application/asgi.py b/application/asgi.py
new file mode 100644
index 00000000..19f8cbe7
--- /dev/null
+++ b/application/asgi.py
@@ -0,0 +1,33 @@
+"""ASGI entrypoint: Flask (WSGI) + FastMCP on the same process."""
+
+from __future__ import annotations
+
+from a2wsgi import WSGIMiddleware
+from starlette.applications import Starlette
+from starlette.middleware import Middleware
+from starlette.middleware.cors import CORSMiddleware
+from starlette.routing import Mount
+
+from application.app import app as flask_app
+from application.mcp_server import mcp
+
+_WSGI_THREADPOOL = 32
+
+mcp_app = mcp.http_app(path="/")
+
+asgi_app = Starlette(
+    routes=[
+        Mount("/mcp", app=mcp_app),
+        Mount("/", app=WSGIMiddleware(flask_app, workers=_WSGI_THREADPOOL)),
+    ],
+    middleware=[
+        Middleware(
+            CORSMiddleware,
+            allow_origins=["*"],
+            allow_methods=["GET", "POST", "PUT", "DELETE", "OPTIONS"],
+            allow_headers=["Content-Type", "Authorization", "Mcp-Session-Id"],
+            expose_headers=["Mcp-Session-Id"],
+        ),
+    ],
+    lifespan=mcp_app.lifespan,
+)
diff --git a/application/gunicorn_conf.py b/application/gunicorn_conf.py
new file mode 100644
index 00000000..4b56dc8e
--- /dev/null
+++ b/application/gunicorn_conf.py
@@ -0,0 +1,72 @@
+"""Gunicorn config — keeps uvicorn's access log in NCSA format."""
+
+from __future__ import annotations
+
+import logging
+import logging.config
+
+# NCSA common log format:
+#   %(h)s %(l)s %(u)s %(t)s "%(r)s" %(s)s %(b)s "%(f)s" "%(a)s"
+# Uvicorn's access formatter exposes a ``client_addr``/``request_line``/
+# ``status_code`` trio but not the full NCSA field set, so we re-derive
+# what we can.
+_NCSA_FMT = (
+    '%(client_addr)s - - [%(asctime)s] "%(request_line)s" %(status_code)s'
+)
+
+logconfig_dict = {
+    "version": 1,
+    "disable_existing_loggers": False,
+    "formatters": {
+        "ncsa_access": {
+            "()": "uvicorn.logging.AccessFormatter",
+            "fmt": _NCSA_FMT,
+            "datefmt": "%d/%b/%Y:%H:%M:%S %z",
+            "use_colors": False,
+        },
+        "default": {
+            "format": "[%(asctime)s] [%(process)d] [%(levelname)s] %(name)s: %(message)s",
+        },
+    },
+    "handlers": {
+        "access": {
+            "class": "logging.StreamHandler",
+            "formatter": "ncsa_access",
+            "stream": "ext://sys.stdout",
+        },
+        "default": {
+            "class": "logging.StreamHandler",
+            "formatter": "default",
+            "stream": "ext://sys.stderr",
+        },
+    },
+    "loggers": {
+        "uvicorn": {"handlers": ["default"], "level": "INFO", "propagate": False},
+        "uvicorn.error": {
+            "handlers": ["default"],
+            "level": "INFO",
+            "propagate": False,
+        },
+        "uvicorn.access": {
+            "handlers": ["access"],
+            "level": "INFO",
+            "propagate": False,
+        },
+        "gunicorn.error": {
+            "handlers": ["default"],
+            "level": "INFO",
+            "propagate": False,
+        },
+        "gunicorn.access": {
+            "handlers": ["access"],
+            "level": "INFO",
+            "propagate": False,
+        },
+    },
+    "root": {"handlers": ["default"], "level": "INFO"},
+}
+
+
+def on_starting(server):  # pragma: no cover — gunicorn hook
+    """Ensure gunicorn's own loggers use the configured handlers."""
+    logging.config.dictConfig(logconfig_dict)
diff --git a/application/mcp_server.py b/application/mcp_server.py
new file mode 100644
index 00000000..23b074f1
--- /dev/null
+++ b/application/mcp_server.py
@@ -0,0 +1,59 @@
+"""FastMCP server exposing DocsGPT retrieval over streamable HTTP.
+
+Mounted at ``/mcp`` by ``application/asgi.py``. Bearer tokens are the
+existing DocsGPT agent API keys — no new credential surface.
+
+The tool reads the ``Authorization`` header directly via
+``get_http_headers(include={"authorization"})``. The ``include`` kwarg
+is required: by default ``get_http_headers`` strips ``authorization``
+(and a handful of other hop-by-hop headers) so they aren't forwarded
+to downstream services — since we deliberately want the caller's
+token, we opt it back in.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import logging
+
+from fastmcp import FastMCP
+from fastmcp.server.dependencies import get_http_headers
+
+from application.services.search_service import (
+    InvalidAPIKey,
+    SearchFailed,
+    search,
+)
+
+logger = logging.getLogger(__name__)
+
+mcp = FastMCP("docsgpt")
+
+
+def _extract_bearer_token() -> str | None:
+    auth = get_http_headers(include={"authorization"}).get("authorization", "")
+    parts = auth.split(None, 1)
+    if len(parts) != 2 or parts[0].lower() != "bearer" or not parts[1]:
+        return None
+    return parts[1]
+
+
+@mcp.tool
+async def search_docs(query: str, chunks: int = 5) -> list[dict]:
+    """Search the caller's DocsGPT knowledge base.
+
+    Authentication is via ``Authorization: Bearer <agent-api-key>`` on
+    the MCP request — the same opaque key that ``/api/search`` accepts
+    in its JSON body. Returns at most ``chunks`` hits, each a dict with
+    ``text``, ``title``, ``source`` keys.
+    """
+    api_key = _extract_bearer_token()
+    if not api_key:
+        raise PermissionError("Missing Bearer token")
+    try:
+        return await asyncio.to_thread(search, api_key, query, chunks)
+    except InvalidAPIKey as exc:
+        raise PermissionError("Invalid API key") from exc
+    except SearchFailed:
+        logger.exception("search_docs failed")
+        raise
diff --git a/application/requirements.txt b/application/requirements.txt
index 2ff38b00..57597811 100644
--- a/application/requirements.txt
+++ b/application/requirements.txt
@@ -1,5 +1,7 @@
+a2wsgi==1.10.10
 alembic>=1.13,<2
 anthropic==0.88.0
+asgiref>=3.11.1
 boto3==1.42.83
 beautifulsoup4==4.14.3
 cel-python==0.5.0
@@ -14,7 +16,7 @@ docx2txt==0.9
 ddgs>=8.0.0
 fast-ebook
 elevenlabs==2.43.0
-Flask==3.1.3
+Flask==3.1.1
 faiss-cpu==1.13.2
 fastmcp==3.2.4
 flask-restx==1.3.2
@@ -76,6 +78,7 @@ requests==2.33.1
 retry==0.9.2
 sentence-transformers==5.3.0
 sqlalchemy>=2.0,<3
+starlette>=1.0,<2
 tiktoken==0.12.0
 tokenizers==0.22.2
 torch==2.11.0
@@ -85,6 +88,8 @@ typing-extensions==4.15.0
 typing-inspect==0.9.0
 tzdata==2026.1
 urllib3==2.6.3
+uvicorn[standard]>=0.30,<1
+uvicorn-worker>=0.4,<1
 vine==5.1.0
 wcwidth==0.6.0
 werkzeug>=3.1.0
diff --git a/application/services/__init__.py b/application/services/__init__.py
new file mode 100644
index 00000000..e69de29b
diff --git a/application/services/search_service.py b/application/services/search_service.py
new file mode 100644
index 00000000..8febd79b
--- /dev/null
+++ b/application/services/search_service.py
@@ -0,0 +1,153 @@
+"""Shared retrieval service used by the HTTP search route and the MCP tool.
+
+Flask-free. Raises domain exceptions (``InvalidAPIKey``, ``SearchFailed``)
+that callers translate into their own wire protocol (HTTP status codes,
+MCP error responses, etc.).
+"""
+
+from __future__ import annotations
+
+import logging
+from typing import Any, Dict, List
+
+from application.core.settings import settings
+from application.storage.db.repositories.agents import AgentsRepository
+from application.storage.db.session import db_readonly
+from application.vectorstore.vector_creator import VectorCreator
+
+logger = logging.getLogger(__name__)
+
+
+class InvalidAPIKey(Exception):
+    """The supplied ``api_key`` does not resolve to an agent."""
+
+
+class SearchFailed(Exception):
+    """Unexpected error during retrieval (e.g. DB outage). Caller maps to 5xx."""
+
+
+def _collect_source_ids(agent: Dict[str, Any]) -> List[str]:
+    """Extract the ordered list of source UUIDs to search.
+
+    Prefers ``extra_source_ids`` (PG ARRAY(UUID) of multi-source agents);
+    falls back to the legacy single ``source_id`` field.
+    """
+    source_ids: List[str] = []
+    extra = agent.get("extra_source_ids") or []
+    for src in extra:
+        if src:
+            source_ids.append(str(src))
+    if not source_ids:
+        single = agent.get("source_id")
+        if single:
+            source_ids.append(str(single))
+    return source_ids
+
+
+def _search_sources(
+    query: str, source_ids: List[str], chunks: int
+) -> List[Dict[str, Any]]:
+    """Search across each source's vectorstore and return up to ``chunks`` hits.
+
+    Per-source errors are logged and skipped so one broken index doesn't
+    take down the whole search. Results are de-duplicated by content hash.
+    """
+    if chunks <= 0 or not source_ids:
+        return []
+
+    results: List[Dict[str, Any]] = []
+    chunks_per_source = max(1, chunks // len(source_ids))
+    seen_texts: set[int] = set()
+
+    for source_id in source_ids:
+        if not source_id or not source_id.strip():
+            continue
+
+        try:
+            docsearch = VectorCreator.create_vectorstore(
+                settings.VECTOR_STORE, source_id, settings.EMBEDDINGS_KEY
+            )
+            docs = docsearch.search(query, k=chunks_per_source * 2)
+
+            for doc in docs:
+                if len(results) >= chunks:
+                    break
+
+                if hasattr(doc, "page_content") and hasattr(doc, "metadata"):
+                    page_content = doc.page_content
+                    metadata = doc.metadata
+                else:
+                    page_content = doc.get("text", doc.get("page_content", ""))
+                    metadata = doc.get("metadata", {})
+
+                text_hash = hash(page_content[:200])
+                if text_hash in seen_texts:
+                    continue
+                seen_texts.add(text_hash)
+
+                title = metadata.get("title", metadata.get("post_title", ""))
+                if not isinstance(title, str):
+                    title = str(title) if title else ""
+
+                if title:
+                    title = title.split("/")[-1]
+                else:
+                    title = metadata.get("filename", page_content[:50] + "...")
+
+                source = metadata.get("source", source_id)
+
+                results.append(
+                    {
+                        "text": page_content,
+                        "title": title,
+                        "source": source,
+                    }
+                )
+
+            if len(results) >= chunks:
+                break
+
+        except Exception as e:
+            logger.error(
+                f"Error searching vectorstore {source_id}: {e}",
+                exc_info=True,
+            )
+            continue
+
+    return results[:chunks]
+
+
+def search(api_key: str, query: str, chunks: int = 5) -> List[Dict[str, Any]]:
+    """Resolve an agent by API key and search its sources.
+
+    Args:
+        api_key: Agent API key (the opaque string stored on
+            ``agents.key`` in Postgres).
+        query: Free-text search query.
+        chunks: Max number of hits to return.
+
+    Returns:
+        List of hit dicts with ``text``, ``title``, ``source`` keys.
+        Empty list if the agent has no sources configured.
+
+    Raises:
+        InvalidAPIKey: if ``api_key`` does not resolve to an agent.
+        SearchFailed: on unexpected DB / infrastructure errors.
+    """
+    if chunks <= 0:
+        return []
+
+    try:
+        with db_readonly() as conn:
+            agent = AgentsRepository(conn).find_by_key(api_key)
+    except Exception as e:
+        raise SearchFailed("agent lookup failed") from e
+
+    if not agent:
+        raise InvalidAPIKey()
+
+    source_ids = _collect_source_ids(agent)
+    if not source_ids:
+        return []
+
+    return _search_sources(query, source_ids, chunks)
diff --git a/docs/content/Deploying/Development-Environment.mdx b/docs/content/Deploying/Development-Environment.mdx
index 057ed4f1..bbe1ebc6 100644
--- a/docs/content/Deploying/Development-Environment.mdx
+++ b/docs/content/Deploying/Development-Environment.mdx
@@ -104,7 +104,15 @@ To run the DocsGPT backend locally, you'll need to set up a Python environment a
     flask --app application/app.py run --host=0.0.0.0 --port=7091
     ```
 
-    This command will launch the backend server, making it accessible on `http://localhost:7091`.
+    This command will launch the backend server, making it accessible on `http://localhost:7091`. It's the fastest inner-loop option for day-to-day development — the Werkzeug interactive debugger still works and it hot-reloads on source changes. It serves the Flask routes only.
+
+    If you need to exercise the full ASGI stack — the `/mcp` endpoint (FastMCP server), or to match the production runtime — run the ASGI composition under uvicorn instead:
+
+    ```bash
+    uvicorn application.asgi:asgi_app --host 0.0.0.0 --port 7091 --reload
+    ```
+
+    Production uses `gunicorn -k uvicorn_worker.UvicornWorker` against the same `application.asgi:asgi_app` target.
 
 6.  **Start the Celery Worker:**
 
diff --git a/pytest.ini b/pytest.ini
index bfe52165..860b86a7 100644
--- a/pytest.ini
+++ b/pytest.ini
@@ -16,6 +16,7 @@ markers =
     unit: Unit tests
     integration: Integration tests
     slow: Slow running tests
+asyncio_mode = strict
 filterwarnings =
     ignore::DeprecationWarning
     ignore::PendingDeprecationWarning
diff --git a/scripts/mock_llm.py b/scripts/mock_llm.py
new file mode 100644
index 00000000..8ce556ee
--- /dev/null
+++ b/scripts/mock_llm.py
@@ -0,0 +1,137 @@
+"""Mock OpenAI-compatible LLM server for benchmarking.
+
+Fixed 5-second generation (100 tokens × 50 ms/token). No auth. Emits SSE
+chunks in OpenAI's chat.completions streaming format, or a single response
+when stream=false. Run on 127.0.0.1:8090 — point DocsGPT at it via
+OPENAI_BASE_URL=http://127.0.0.1:8090/v1.
+"""
+
+import asyncio
+import json
+import logging
+import time
+import uuid
+
+from fastapi import FastAPI, Request
+from fastapi.responses import JSONResponse, StreamingResponse
+
+TOKEN_COUNT = 100
+TOKEN_DELAY_S = 0.05  # 100 * 0.05 = 5.0 s
+
+logger = logging.getLogger("mock_llm")
+logging.basicConfig(level=logging.INFO, format="%(asctime)s mock: %(message)s")
+
+FILLER_TOKENS = [
+    "Lorem", " ipsum", " dolor", " sit", " amet", ",", " consectetur",
+    " adipiscing", " elit", ".", " Sed", " do", " eiusmod", " tempor",
+    " incididunt", " ut", " labore", " et", " dolore", " magna", " aliqua",
+    ".", " Ut", " enim", " ad", " minim", " veniam", ",", " quis", " nostrud",
+    " exercitation", " ullamco", " laboris", " nisi", " ut", " aliquip",
+    " ex", " ea", " commodo", " consequat", ".", " Duis", " aute", " irure",
+    " dolor", " in", " reprehenderit", " in", " voluptate", " velit",
+    " esse", " cillum", " dolore", " eu", " fugiat", " nulla", " pariatur",
+    ".", " Excepteur", " sint", " occaecat", " cupidatat", " non", " proident",
+    ",", " sunt", " in", " culpa", " qui", " officia", " deserunt",
+    " mollit", " anim", " id", " est", " laborum", ".", " Curabitur",
+    " pretium", " tincidunt", " lacus", ".", " Nulla", " gravida", " orci",
+    " a", " odio", ".", " Nullam", " varius", ",", " turpis", " et",
+    " commodo", " pharetra", ",", " est", " eros", " bibendum", " elit",
+    ".",
+]
+
+app = FastAPI()
+
+
+def _token_stream_id() -> str:
+    return f"chatcmpl-mock-{uuid.uuid4().hex[:12]}"
+
+
+def _sse_chunk(completion_id: str, model: str, delta: dict, finish_reason=None) -> str:
+    payload = {
+        "id": completion_id,
+        "object": "chat.completion.chunk",
+        "created": int(time.time()),
+        "model": model,
+        "choices": [
+            {
+                "index": 0,
+                "delta": delta,
+                "finish_reason": finish_reason,
+            }
+        ],
+    }
+    return f"data: {json.dumps(payload)}\n\n"
+
+
+async def _stream_response(model: str, req_id: str):
+    completion_id = _token_stream_id()
+    yield _sse_chunk(completion_id, model, {"role": "assistant", "content": ""})
+    for i, tok in enumerate(FILLER_TOKENS[:TOKEN_COUNT]):
+        await asyncio.sleep(TOKEN_DELAY_S)
+        yield _sse_chunk(completion_id, model, {"content": tok})
+    yield _sse_chunk(completion_id, model, {}, finish_reason="stop")
+    yield "data: [DONE]\n\n"
+    logger.info("[%s] stream done", req_id)
+
+
+@app.post("/v1/chat/completions")
+async def chat_completions(request: Request):
+    body = await request.json()
+    model = body.get("model", "mock")
+    stream = bool(body.get("stream", False))
+    req_id = uuid.uuid4().hex[:8]
+    logger.info("[%s] /chat/completions stream=%s model=%s max_tokens=%s", req_id, stream, model, body.get("max_tokens"))
+
+    if stream:
+        return StreamingResponse(
+            _stream_response(model, req_id),
+            media_type="text/event-stream",
+            headers={
+                "Cache-Control": "no-cache, no-transform",
+                "X-Accel-Buffering": "no",
+            },
+        )
+
+    await asyncio.sleep(TOKEN_COUNT * TOKEN_DELAY_S)
+    logger.info("[%s] non-stream done", req_id)
+    text = "".join(FILLER_TOKENS[:TOKEN_COUNT])
+    completion_id = _token_stream_id()
+    return JSONResponse(
+        {
+            "id": completion_id,
+            "object": "chat.completion",
+            "created": int(time.time()),
+            "model": model,
+            "choices": [
+                {
+                    "index": 0,
+                    "message": {"role": "assistant", "content": text},
+                    "finish_reason": "stop",
+                }
+            ],
+            "usage": {
+                "prompt_tokens": 10,
+                "completion_tokens": TOKEN_COUNT,
+                "total_tokens": 10 + TOKEN_COUNT,
+            },
+        }
+    )
+
+
+@app.get("/v1/models")
+async def list_models():
+    return {
+        "object": "list",
+        "data": [{"id": "mock", "object": "model", "owned_by": "mock"}],
+    }
+
+
+@app.get("/health")
+async def health():
+    return {"status": "ok"}
+
+
+if __name__ == "__main__":
+    import uvicorn
+
+    uvicorn.run(app, host="127.0.0.1", port=8090, log_level="info")
diff --git a/tests/api/answer/routes/test_search.py b/tests/api/answer/routes/test_search.py
index 6f38934c..73bda7dc 100644
--- a/tests/api/answer/routes/test_search.py
+++ b/tests/api/answer/routes/test_search.py
@@ -1,3 +1,17 @@
+"""Tests for /api/search route (application/api/answer/routes/search.py).
+
+Retrieval logic lives in ``application/services/search_service.py`` and
+has its own unit tests in ``tests/services/test_search_service.py``. The
+tests below focus on what the route specifically owns:
+
+* Request validation (400 for missing fields).
+* Translation of the service's ``InvalidAPIKey`` / ``SearchFailed``
+  exceptions to HTTP status codes (401 / 500).
+* End-to-end happy path against a real ephemeral Postgres via
+  ``pg_conn``, to catch regressions in the route's wiring to the
+  service and repositories.
+"""
+
 from contextlib import contextmanager
 from unittest.mock import MagicMock, patch
 
@@ -6,254 +20,97 @@ import pytest
 
 @pytest.mark.unit
 class TestSearchResourceValidation:
-    pass
-
-    def test_returns_error_when_question_missing(self, mock_mongo_db, flask_app):
+    def test_returns_400_when_question_missing(self, flask_app):
         from application.api.answer.routes.search import SearchResource
 
         with flask_app.app_context():
-            with flask_app.test_request_context(
-                json={"api_key": "test_key"}
-            ):
-                resource = SearchResource()
-                result = resource.post()
-
+            with flask_app.test_request_context(json={"api_key": "test_key"}):
+                result = SearchResource().post()
                 assert result.status_code == 400
                 assert "question" in result.json["error"]
 
-    def test_returns_error_when_api_key_missing(self, mock_mongo_db, flask_app):
+    def test_returns_400_when_api_key_missing(self, flask_app):
         from application.api.answer.routes.search import SearchResource
 
         with flask_app.app_context():
-            with flask_app.test_request_context(
-                json={"question": "test query"}
-            ):
-                resource = SearchResource()
-                result = resource.post()
-
+            with flask_app.test_request_context(json={"question": "test query"}):
+                result = SearchResource().post()
                 assert result.status_code == 400
                 assert "api_key" in result.json["error"]
 
 
-
 @pytest.mark.unit
-class TestGetSourcesFromApiKey:
-    pass
+class TestSearchResourceExceptionMapping:
+    """Verify the route maps service exceptions to HTTP status codes.
 
-    def test_returns_source_id_via_patched_method(self, mock_mongo_db, flask_app):
-        """Test that _get_sources_from_api_key can return multiple sources via patch."""
+    The service function itself is patched; these tests do not care about
+    the search logic — only that 401/500/200 are produced correctly from
+    the three possible service outcomes.
+    """
+
+    def test_invalid_api_key_returns_401(self, flask_app):
+        from application.api.answer.routes.search import SearchResource
+        from application.services.search_service import InvalidAPIKey
+
+        with flask_app.app_context(), flask_app.test_request_context(
+            json={"question": "q", "api_key": "bad"}
+        ), patch(
+            "application.api.answer.routes.search.search",
+            side_effect=InvalidAPIKey(),
+        ):
+            result = SearchResource().post()
+        assert result.status_code == 401
+        assert result.json == {"error": "Invalid API key"}
+
+    def test_search_failed_returns_500(self, flask_app):
+        from application.api.answer.routes.search import SearchResource
+        from application.services.search_service import SearchFailed
+
+        with flask_app.app_context(), flask_app.test_request_context(
+            json={"question": "q", "api_key": "k"}
+        ), patch(
+            "application.api.answer.routes.search.search",
+            side_effect=SearchFailed("boom"),
+        ):
+            result = SearchResource().post()
+        assert result.status_code == 500
+        assert result.json == {"error": "Search failed"}
+
+    def test_happy_path_passes_service_result_through(self, flask_app):
         from application.api.answer.routes.search import SearchResource
 
-        with flask_app.app_context():
-            resource = SearchResource()
+        hits = [{"text": "t", "title": "T", "source": "s"}]
+        with flask_app.app_context(), flask_app.test_request_context(
+            json={"question": "q", "api_key": "k", "chunks": 7}
+        ), patch(
+            "application.api.answer.routes.search.search",
+            return_value=hits,
+        ) as mock_search:
+            result = SearchResource().post()
+        assert result.status_code == 200
+        assert result.json == hits
+        mock_search.assert_called_once_with("k", "q", 7)
 
-            with patch.object(resource, "_get_sources_from_api_key", return_value=["src1", "src2"]):
-                result = resource._get_sources_from_api_key("any_key")
-
-            assert len(result) == 2
-            assert "src1" in result
-            assert "src2" in result
-
-
-
-@pytest.mark.unit
-class TestSearchVectorstores:
-    pass
-
-    def test_returns_empty_when_no_source_ids(self, mock_mongo_db, flask_app):
+    def test_default_chunks_is_5(self, flask_app):
         from application.api.answer.routes.search import SearchResource
 
-        with flask_app.app_context():
-            resource = SearchResource()
-
-            result = resource._search_vectorstores("test query", [], 5)
-
-            assert result == []
-
-    def test_skips_empty_source_ids(self, mock_mongo_db, flask_app):
-        from application.api.answer.routes.search import SearchResource
-
-        with flask_app.app_context():
-            resource = SearchResource()
-
-            with patch(
-                "application.api.answer.routes.search.VectorCreator.create_vectorstore"
-            ) as mock_create:
-                mock_vectorstore = MagicMock()
-                mock_vectorstore.search.return_value = []
-                mock_create.return_value = mock_vectorstore
-
-                result = resource._search_vectorstores("test query", ["", "  "], 5)
-
-                mock_create.assert_not_called()
-                assert result == []
-
-    def test_returns_search_results(self, mock_mongo_db, flask_app):
-        from application.api.answer.routes.search import SearchResource
-
-        with flask_app.app_context():
-            resource = SearchResource()
-
-            mock_doc = {
-                "text": "Test content",
-                "page_content": "Test content",
-                "metadata": {
-                    "title": "Test Title",
-                    "source": "/path/to/doc",
-                },
-            }
-
-            with patch(
-                "application.api.answer.routes.search.VectorCreator.create_vectorstore"
-            ) as mock_create:
-                mock_vectorstore = MagicMock()
-                mock_vectorstore.search.return_value = [mock_doc]
-                mock_create.return_value = mock_vectorstore
-
-                result = resource._search_vectorstores("test query", ["source_id"], 5)
-
-                assert len(result) == 1
-                assert result[0]["text"] == "Test content"
-                assert result[0]["title"] == "Test Title"
-                assert result[0]["source"] == "/path/to/doc"
-
-    def test_handles_langchain_document_format(self, mock_mongo_db, flask_app):
-        from application.api.answer.routes.search import SearchResource
-
-        with flask_app.app_context():
-            resource = SearchResource()
-
-            mock_doc = MagicMock()
-            mock_doc.page_content = "Langchain content"
-            mock_doc.metadata = {"title": "LC Title", "source": "/lc/path"}
-
-            with patch(
-                "application.api.answer.routes.search.VectorCreator.create_vectorstore"
-            ) as mock_create:
-                mock_vectorstore = MagicMock()
-                mock_vectorstore.search.return_value = [mock_doc]
-                mock_create.return_value = mock_vectorstore
-
-                result = resource._search_vectorstores("test query", ["source_id"], 5)
-
-                assert len(result) == 1
-                assert result[0]["text"] == "Langchain content"
-                assert result[0]["title"] == "LC Title"
-
-    def test_respects_chunks_limit(self, mock_mongo_db, flask_app):
-        from application.api.answer.routes.search import SearchResource
-
-        with flask_app.app_context():
-            resource = SearchResource()
-
-            mock_docs = [
-                {"text": f"Content {i}", "metadata": {"title": f"Title {i}"}}
-                for i in range(10)
-            ]
-
-            with patch(
-                "application.api.answer.routes.search.VectorCreator.create_vectorstore"
-            ) as mock_create:
-                mock_vectorstore = MagicMock()
-                mock_vectorstore.search.return_value = mock_docs
-                mock_create.return_value = mock_vectorstore
-
-                result = resource._search_vectorstores("test query", ["source_id"], 3)
-
-                assert len(result) == 3
-
-    def test_deduplicates_results(self, mock_mongo_db, flask_app):
-        from application.api.answer.routes.search import SearchResource
-
-        with flask_app.app_context():
-            resource = SearchResource()
-
-            duplicate_text = "Duplicate content " * 20
-            mock_docs = [
-                {"text": duplicate_text, "metadata": {"title": "Title 1"}},
-                {"text": duplicate_text, "metadata": {"title": "Title 2"}},
-                {"text": "Unique content", "metadata": {"title": "Title 3"}},
-            ]
-
-            with patch(
-                "application.api.answer.routes.search.VectorCreator.create_vectorstore"
-            ) as mock_create:
-                mock_vectorstore = MagicMock()
-                mock_vectorstore.search.return_value = mock_docs
-                mock_create.return_value = mock_vectorstore
-
-                result = resource._search_vectorstores("test query", ["source_id"], 5)
-
-                assert len(result) == 2
-
-    def test_handles_vectorstore_error_gracefully(self, mock_mongo_db, flask_app):
-        from application.api.answer.routes.search import SearchResource
-
-        with flask_app.app_context():
-            resource = SearchResource()
-
-            with patch(
-                "application.api.answer.routes.search.VectorCreator.create_vectorstore"
-            ) as mock_create:
-                mock_create.side_effect = Exception("Vectorstore error")
-
-                result = resource._search_vectorstores("test query", ["source_id"], 5)
-
-                assert result == []
-
-    def test_uses_filename_as_title_fallback(self, mock_mongo_db, flask_app):
-        from application.api.answer.routes.search import SearchResource
-
-        with flask_app.app_context():
-            resource = SearchResource()
-
-            mock_doc = {
-                "text": "Content without title",
-                "metadata": {"filename": "document.pdf"},
-            }
-
-            with patch(
-                "application.api.answer.routes.search.VectorCreator.create_vectorstore"
-            ) as mock_create:
-                mock_vectorstore = MagicMock()
-                mock_vectorstore.search.return_value = [mock_doc]
-                mock_create.return_value = mock_vectorstore
-
-                result = resource._search_vectorstores("test query", ["source_id"], 5)
-
-                assert result[0]["title"] == "document.pdf"
-
-    def test_uses_content_snippet_as_title_last_resort(self, mock_mongo_db, flask_app):
-        from application.api.answer.routes.search import SearchResource
-
-        with flask_app.app_context():
-            resource = SearchResource()
-
-            mock_doc = {
-                "text": "Content without any title metadata at all",
-                "metadata": {},
-            }
-
-            with patch(
-                "application.api.answer.routes.search.VectorCreator.create_vectorstore"
-            ) as mock_create:
-                mock_vectorstore = MagicMock()
-                mock_vectorstore.search.return_value = [mock_doc]
-                mock_create.return_value = mock_vectorstore
-
-                result = resource._search_vectorstores("test query", ["source_id"], 5)
-
-                assert "Content without any title" in result[0]["title"]
-                assert result[0]["title"].endswith("...")
-
-
-@pytest.mark.unit
-class TestSearchEndpoint:
-    pass
+        with flask_app.app_context(), flask_app.test_request_context(
+            json={"question": "q", "api_key": "k"}  # no chunks field
+        ), patch(
+            "application.api.answer.routes.search.search",
+            return_value=[],
+        ) as mock_search:
+            SearchResource().post()
+        mock_search.assert_called_once_with("k", "q", 5)
 
 
 # ---------------------------------------------------------------------------
-# Real-PG tests for SearchResource.
+# End-to-end against a real ephemeral Postgres.
+#
+# These exercise the full route → service → repository → DB path, patching
+# only ``VectorCreator.create_vectorstore`` (so we don't need real embeddings
+# or a vector index). ``db_readonly`` is redirected at the *service* module
+# since that's where the import now lives.
 # ---------------------------------------------------------------------------
 
 
@@ -264,7 +121,7 @@ def _patch_search_db(conn):
         yield conn
 
     with patch(
-        "application.api.answer.routes.search.db_readonly", _yield
+        "application.services.search_service.db_readonly", _yield
     ):
         yield
 
@@ -298,9 +155,7 @@ class TestSearchResourcePgConn:
     def test_search_returns_results(self, pg_conn, flask_app):
         from application.api.answer.routes.search import SearchResource
         from application.storage.db.repositories.agents import AgentsRepository
-        from application.storage.db.repositories.sources import (
-            SourcesRepository,
-        )
+        from application.storage.db.repositories.sources import SourcesRepository
 
         src = SourcesRepository(pg_conn).create("src", user_id="u")
         AgentsRepository(pg_conn).create(
@@ -315,7 +170,7 @@ class TestSearchResourcePgConn:
         ]
 
         with _patch_search_db(pg_conn), patch(
-            "application.api.answer.routes.search.VectorCreator.create_vectorstore",
+            "application.services.search_service.VectorCreator.create_vectorstore",
             return_value=fake_vs,
         ), flask_app.app_context():
             with flask_app.test_request_context(
@@ -328,9 +183,7 @@ class TestSearchResourcePgConn:
     def test_search_uses_extra_source_ids(self, pg_conn, flask_app):
         from application.api.answer.routes.search import SearchResource
         from application.storage.db.repositories.agents import AgentsRepository
-        from application.storage.db.repositories.sources import (
-            SourcesRepository,
-        )
+        from application.storage.db.repositories.sources import SourcesRepository
 
         src1 = SourcesRepository(pg_conn).create("s1", user_id="u")
         src2 = SourcesRepository(pg_conn).create("s2", user_id="u")
@@ -345,7 +198,7 @@ class TestSearchResourcePgConn:
             {"text": "one", "metadata": {"title": "A"}},
         ]
         with _patch_search_db(pg_conn), patch(
-            "application.api.answer.routes.search.VectorCreator.create_vectorstore",
+            "application.services.search_service.VectorCreator.create_vectorstore",
             return_value=fake_vs,
         ), flask_app.app_context():
             with flask_app.test_request_context(
@@ -353,71 +206,3 @@ class TestSearchResourcePgConn:
             ):
                 result = SearchResource().post()
         assert result.status_code == 200
-
-    def test_search_exception_returns_500(self, pg_conn, flask_app):
-        from application.api.answer.routes.search import SearchResource
-        from application.storage.db.repositories.agents import AgentsRepository
-        from application.storage.db.repositories.sources import (
-            SourcesRepository,
-        )
-
-        src = SourcesRepository(pg_conn).create("src", user_id="u")
-        AgentsRepository(pg_conn).create(
-            "u", "a", "published",
-            key="err-key",
-            source_id=str(src["id"]),
-        )
-
-        with _patch_search_db(pg_conn), patch(
-            "application.api.answer.routes.search.SearchResource._get_sources_from_api_key",
-            side_effect=RuntimeError("boom"),
-        ), flask_app.app_context():
-            with flask_app.test_request_context(
-                json={"question": "q", "api_key": "err-key"},
-            ):
-                result = SearchResource().post()
-        assert result.status_code == 500
-
-
-class TestGetSourcesFromApiKeyPg:
-    def test_empty_for_unknown_key(self, pg_conn, flask_app):
-        from application.api.answer.routes.search import SearchResource
-
-        with _patch_search_db(pg_conn), flask_app.app_context():
-            got = SearchResource()._get_sources_from_api_key("nope")
-        assert got == []
-
-    def test_returns_extra_source_ids(self, pg_conn, flask_app):
-        from application.api.answer.routes.search import SearchResource
-        from application.storage.db.repositories.agents import AgentsRepository
-        from application.storage.db.repositories.sources import (
-            SourcesRepository,
-        )
-
-        src = SourcesRepository(pg_conn).create("s", user_id="u")
-        AgentsRepository(pg_conn).create(
-            "u", "a", "published",
-            key="sources-key",
-            extra_source_ids=[str(src["id"])],
-        )
-        with _patch_search_db(pg_conn), flask_app.app_context():
-            got = SearchResource()._get_sources_from_api_key("sources-key")
-        assert got == [str(src["id"])]
-
-    def test_falls_back_to_single_source(self, pg_conn, flask_app):
-        from application.api.answer.routes.search import SearchResource
-        from application.storage.db.repositories.agents import AgentsRepository
-        from application.storage.db.repositories.sources import (
-            SourcesRepository,
-        )
-
-        src = SourcesRepository(pg_conn).create("s", user_id="u")
-        AgentsRepository(pg_conn).create(
-            "u", "a", "published",
-            key="single-key",
-            source_id=str(src["id"]),
-        )
-        with _patch_search_db(pg_conn), flask_app.app_context():
-            got = SearchResource()._get_sources_from_api_key("single-key")
-        assert got == [str(src["id"])]
-
diff --git a/tests/requirements.txt b/tests/requirements.txt
index b555cbf8..2af47c93 100644
--- a/tests/requirements.txt
+++ b/tests/requirements.txt
@@ -1,4 +1,5 @@
 pytest>=8.0.0
+pytest-asyncio>=0.23
 pytest-cov>=4.1.0
 coverage>=7.4.0
 pytest-postgresql>=6.0.0
diff --git a/tests/services/__init__.py b/tests/services/__init__.py
new file mode 100644
index 00000000..e69de29b
diff --git a/tests/services/test_mcp_server.py b/tests/services/test_mcp_server.py
new file mode 100644
index 00000000..c1da1cbd
--- /dev/null
+++ b/tests/services/test_mcp_server.py
@@ -0,0 +1,134 @@
+"""Tests for application/mcp_server.py.
+
+The server module exposes one FastMCP tool, ``search_docs``, that reads
+the caller's ``Authorization: Bearer <key>`` header via
+``get_http_headers()`` and delegates to
+``application.services.search_service.search``. These tests exercise
+the tool directly by patching ``get_http_headers`` and ``search``; the
+full HTTP-layer plumbing (mount, lifespan, session handshake) is
+covered by ``tests/test_asgi.py``.
+"""
+
+from unittest.mock import patch
+
+import pytest
+
+
+@pytest.mark.unit
+class TestSearchDocsTool:
+    @pytest.mark.asyncio
+    async def test_missing_bearer_raises_permission_error(self):
+        from application.mcp_server import search_docs
+
+        with patch(
+            "application.mcp_server.get_http_headers", return_value={}
+        ):
+            with pytest.raises(PermissionError):
+                await search_docs(query="hi")
+
+    @pytest.mark.asyncio
+    async def test_non_bearer_header_raises_permission_error(self):
+        from application.mcp_server import search_docs
+
+        with patch(
+            "application.mcp_server.get_http_headers",
+            return_value={"authorization": "Basic dXNlcjpwYXNz"},
+        ):
+            with pytest.raises(PermissionError):
+                await search_docs(query="hi")
+
+    @pytest.mark.asyncio
+    async def test_blank_bearer_token_raises_permission_error(self):
+        from application.mcp_server import search_docs
+
+        with patch(
+            "application.mcp_server.get_http_headers",
+            return_value={"authorization": "Bearer    "},
+        ):
+            with pytest.raises(PermissionError):
+                await search_docs(query="hi")
+
+    @pytest.mark.asyncio
+    async def test_invalid_api_key_raises_permission_error(self):
+        from application.mcp_server import search_docs
+        from application.services.search_service import InvalidAPIKey
+
+        with (
+            patch(
+                "application.mcp_server.get_http_headers",
+                return_value={"authorization": "Bearer bogus"},
+            ),
+            patch(
+                "application.mcp_server.search", side_effect=InvalidAPIKey()
+            ),
+        ):
+            with pytest.raises(PermissionError):
+                await search_docs(query="hi")
+
+    @pytest.mark.asyncio
+    async def test_search_failed_bubbles_up(self):
+        from application.mcp_server import search_docs
+        from application.services.search_service import SearchFailed
+
+        with (
+            patch(
+                "application.mcp_server.get_http_headers",
+                return_value={"authorization": "Bearer k"},
+            ),
+            patch(
+                "application.mcp_server.search",
+                side_effect=SearchFailed("boom"),
+            ),
+        ):
+            with pytest.raises(SearchFailed):
+                await search_docs(query="hi")
+
+    @pytest.mark.asyncio
+    async def test_happy_path_passes_args_and_returns_hits(self):
+        from application.mcp_server import search_docs
+
+        hits = [{"text": "t", "title": "T", "source": "s"}]
+        with (
+            patch(
+                "application.mcp_server.get_http_headers",
+                return_value={"authorization": "Bearer the-key"},
+            ),
+            patch(
+                "application.mcp_server.search", return_value=hits
+            ) as mock_search,
+        ):
+            out = await search_docs(query="q", chunks=7)
+        assert out == hits
+        mock_search.assert_called_once_with("the-key", "q", 7)
+
+    @pytest.mark.asyncio
+    async def test_default_chunks_is_5(self):
+        from application.mcp_server import search_docs
+
+        with (
+            patch(
+                "application.mcp_server.get_http_headers",
+                return_value={"authorization": "Bearer k"},
+            ),
+            patch(
+                "application.mcp_server.search", return_value=[]
+            ) as mock_search,
+        ):
+            await search_docs(query="q")
+        mock_search.assert_called_once_with("k", "q", 5)
+
+    @pytest.mark.asyncio
+    async def test_bearer_scheme_case_insensitive(self):
+        from application.mcp_server import search_docs
+
+        with (
+            patch(
+                "application.mcp_server.get_http_headers",
+                return_value={"authorization": "bearer lowercase-scheme"},
+            ),
+            patch(
+                "application.mcp_server.search", return_value=[]
+            ) as mock_search,
+        ):
+            await search_docs(query="q")
+        mock_search.assert_called_once_with("lowercase-scheme", "q", 5)
diff --git a/tests/services/test_search_service.py b/tests/services/test_search_service.py
new file mode 100644
index 00000000..2b725404
--- /dev/null
+++ b/tests/services/test_search_service.py
@@ -0,0 +1,230 @@
+"""Unit tests for application/services/search_service.py.
+
+Tests exercise the service function in isolation — AgentsRepository is
+stubbed via a patched ``db_readonly`` context manager, and
+``VectorCreator.create_vectorstore`` is patched to return a fake
+vectorstore. No Flask app context, no real DB, no real embeddings.
+"""
+
+from contextlib import contextmanager
+from unittest.mock import MagicMock, patch
+
+import pytest
+
+from application.services.search_service import (
+    InvalidAPIKey,
+    SearchFailed,
+    _collect_source_ids,
+    search,
+)
+
+
+@contextmanager
+def _fake_db_readonly(agent_data):
+    """Patch ``db_readonly`` so ``AgentsRepository.find_by_key`` returns ``agent_data``."""
+    agents_repo = MagicMock()
+    agents_repo.find_by_key.return_value = agent_data
+
+    @contextmanager
+    def _yield_conn():
+        yield MagicMock()
+
+    with patch(
+        "application.services.search_service.db_readonly", _yield_conn
+    ), patch(
+        "application.services.search_service.AgentsRepository",
+        return_value=agents_repo,
+    ):
+        yield
+
+
+@pytest.mark.unit
+class TestCollectSourceIds:
+    def test_empty_when_no_sources(self):
+        assert _collect_source_ids({}) == []
+
+    def test_returns_extra_source_ids(self):
+        agent = {"extra_source_ids": ["s1", "s2"], "source_id": "legacy"}
+        assert _collect_source_ids(agent) == ["s1", "s2"]
+
+    def test_falls_back_to_single_source_id(self):
+        agent = {"extra_source_ids": [], "source_id": "s1"}
+        assert _collect_source_ids(agent) == ["s1"]
+
+    def test_skips_empty_entries_in_extra(self):
+        agent = {"extra_source_ids": ["", None, "s1"], "source_id": "fallback"}
+        assert _collect_source_ids(agent) == ["s1"]
+
+
+@pytest.mark.unit
+class TestSearchInvalidAPIKey:
+    def test_raises_when_key_unknown(self):
+        with _fake_db_readonly(None):
+            with pytest.raises(InvalidAPIKey):
+                search("does-not-exist", "hello", 5)
+
+    def test_raises_search_failed_on_db_error(self):
+        @contextmanager
+        def _yield_conn():
+            yield MagicMock()
+
+        agents_repo = MagicMock()
+        agents_repo.find_by_key.side_effect = RuntimeError("db down")
+
+        with patch(
+            "application.services.search_service.db_readonly", _yield_conn
+        ), patch(
+            "application.services.search_service.AgentsRepository",
+            return_value=agents_repo,
+        ):
+            with pytest.raises(SearchFailed):
+                search("any-key", "hello", 5)
+
+
+@pytest.mark.unit
+class TestSearchEmptyWhenNoSources:
+    def test_returns_empty_when_agent_has_no_sources(self):
+        with _fake_db_readonly({"extra_source_ids": [], "source_id": None}):
+            assert search("k", "q", 5) == []
+
+    def test_returns_empty_for_zero_chunks_without_db_lookup(self):
+        with patch("application.services.search_service.db_readonly") as mock_db:
+            assert search("k", "q", 0) == []
+        mock_db.assert_not_called()
+
+    def test_returns_empty_for_negative_chunks_without_db_lookup(self):
+        with patch("application.services.search_service.db_readonly") as mock_db:
+            assert search("k", "q", -1) == []
+        mock_db.assert_not_called()
+
+
+@pytest.mark.unit
+class TestSearchResults:
+    def test_returns_hit_shape(self):
+        agent = {"source_id": "src-1", "extra_source_ids": []}
+        fake_vs = MagicMock()
+        fake_vs.search.return_value = [
+            {
+                "text": "Test content",
+                "metadata": {"title": "Test Title", "source": "/path/to/doc"},
+            }
+        ]
+        with _fake_db_readonly(agent), patch(
+            "application.services.search_service.VectorCreator.create_vectorstore",
+            return_value=fake_vs,
+        ):
+            results = search("k", "q", 5)
+        assert results == [
+            {"text": "Test content", "title": "Test Title", "source": "/path/to/doc"}
+        ]
+
+    def test_handles_langchain_document_format(self):
+        agent = {"source_id": "src-1", "extra_source_ids": []}
+        lc_doc = MagicMock()
+        lc_doc.page_content = "Langchain content"
+        lc_doc.metadata = {"title": "LC Title", "source": "/lc/path"}
+
+        fake_vs = MagicMock()
+        fake_vs.search.return_value = [lc_doc]
+
+        with _fake_db_readonly(agent), patch(
+            "application.services.search_service.VectorCreator.create_vectorstore",
+            return_value=fake_vs,
+        ):
+            results = search("k", "q", 5)
+        assert len(results) == 1
+        assert results[0]["text"] == "Langchain content"
+        assert results[0]["title"] == "LC Title"
+
+    def test_respects_chunks_cap(self):
+        agent = {"source_id": "src-1", "extra_source_ids": []}
+        docs = [
+            {"text": f"Content {i}", "metadata": {"title": f"T{i}"}}
+            for i in range(10)
+        ]
+        fake_vs = MagicMock()
+        fake_vs.search.return_value = docs
+
+        with _fake_db_readonly(agent), patch(
+            "application.services.search_service.VectorCreator.create_vectorstore",
+            return_value=fake_vs,
+        ):
+            results = search("k", "q", 3)
+        assert len(results) == 3
+
+    def test_deduplicates_results_by_content_prefix(self):
+        agent = {"source_id": "src-1", "extra_source_ids": []}
+        dup_text = "Duplicate content " * 20
+        docs = [
+            {"text": dup_text, "metadata": {"title": "T1"}},
+            {"text": dup_text, "metadata": {"title": "T2"}},
+            {"text": "Unique content", "metadata": {"title": "T3"}},
+        ]
+        fake_vs = MagicMock()
+        fake_vs.search.return_value = docs
+
+        with _fake_db_readonly(agent), patch(
+            "application.services.search_service.VectorCreator.create_vectorstore",
+            return_value=fake_vs,
+        ):
+            results = search("k", "q", 5)
+        assert len(results) == 2
+
+    def test_skips_broken_source_and_returns_from_healthy_ones(self):
+        # Two sources — the first raises, the second returns a doc. The
+        # caller should still get the healthy source's result.
+        agent = {"extra_source_ids": ["broken", "ok"], "source_id": None}
+        healthy_vs = MagicMock()
+        healthy_vs.search.return_value = [
+            {"text": "ok content", "metadata": {"title": "Ok"}}
+        ]
+
+        def create_vs(store, source_id, key):
+            if source_id == "broken":
+                raise RuntimeError("vector index missing")
+            return healthy_vs
+
+        with _fake_db_readonly(agent), patch(
+            "application.services.search_service.VectorCreator.create_vectorstore",
+            side_effect=create_vs,
+        ):
+            results = search("k", "q", 5)
+        assert len(results) == 1
+        assert results[0]["text"] == "ok content"
+
+    def test_uses_filename_when_title_missing(self):
+        agent = {"source_id": "src-1", "extra_source_ids": []}
+        fake_vs = MagicMock()
+        fake_vs.search.return_value = [
+            {"text": "body", "metadata": {"filename": "document.pdf"}}
+        ]
+        with _fake_db_readonly(agent), patch(
+            "application.services.search_service.VectorCreator.create_vectorstore",
+            return_value=fake_vs,
+        ):
+            results = search("k", "q", 5)
+        assert results[0]["title"] == "document.pdf"
+
+    def test_uses_content_snippet_as_title_last_resort(self):
+        agent = {"source_id": "src-1", "extra_source_ids": []}
+        fake_vs = MagicMock()
+        fake_vs.search.return_value = [
+            {"text": "Content without any title metadata at all", "metadata": {}}
+        ]
+        with _fake_db_readonly(agent), patch(
+            "application.services.search_service.VectorCreator.create_vectorstore",
+            return_value=fake_vs,
+        ):
+            results = search("k", "q", 5)
+        assert results[0]["title"].endswith("...")
+        assert "Content without any title" in results[0]["title"]
+
+    def test_skips_empty_source_ids(self):
+        # ``source_id=" "`` only — after strip() this leaves no real source.
+        agent = {"extra_source_ids": ["  ", ""], "source_id": None}
+        with _fake_db_readonly(agent), patch(
+            "application.services.search_service.VectorCreator.create_vectorstore"
+        ) as mock_create:
+            results = search("k", "q", 5)
+        mock_create.assert_not_called()
+        assert results == []
diff --git a/tests/test_app_routes.py b/tests/test_app_routes.py
index eb5fe26f..656b550b 100644
--- a/tests/test_app_routes.py
+++ b/tests/test_app_routes.py
@@ -105,11 +105,26 @@ class TestAuthenticateRequest:
             assert response.status_code == 200
 
 
-class TestAfterRequest:
+class TestFlaskCors:
 
     @pytest.mark.unit
-    def test_cors_headers(self, client):
-        response = client.get("/api/health")
-        assert response.headers.get("Access-Control-Allow-Origin") == "*"
-        assert "Content-Type" in response.headers.get("Access-Control-Allow-Headers", "")
-        assert "GET" in response.headers.get("Access-Control-Allow-Methods", "")
+    def test_cors_headers_on_flask_route(self, client):
+        response = client.get("/api/health", headers={"Origin": "http://localhost:5173"})
+        assert response.headers["Access-Control-Allow-Origin"] == "*"
+        assert response.headers["Access-Control-Allow-Headers"] == "Content-Type, Authorization"
+        assert response.headers["Access-Control-Allow-Methods"] == "GET, POST, PUT, DELETE, OPTIONS"
+
+    @pytest.mark.unit
+    def test_cors_headers_on_flask_preflight(self, client):
+        response = client.options(
+            "/api/health",
+            headers={
+                "Origin": "http://localhost:5173",
+                "Access-Control-Request-Method": "GET",
+                "Access-Control-Request-Headers": "Content-Type",
+            },
+        )
+        assert response.status_code == 200
+        assert response.headers["Access-Control-Allow-Origin"] == "*"
+        assert response.headers["Access-Control-Allow-Headers"] == "Content-Type, Authorization"
+        assert response.headers["Access-Control-Allow-Methods"] == "GET, POST, PUT, DELETE, OPTIONS"
diff --git a/tests/test_asgi.py b/tests/test_asgi.py
new file mode 100644
index 00000000..57fd4fc9
--- /dev/null
+++ b/tests/test_asgi.py
@@ -0,0 +1,136 @@
+"""Smoke tests for application/asgi.py.
+
+The goal isn't to re-test Flask or FastMCP internals — it's to catch
+regressions in the wiring: mounts resolve, CORS headers emit, lifespan
+runs (without it, the /mcp session manager raises "Task group is not
+initialized"), routing to ``/`` vs ``/mcp`` doesn't cross paths.
+
+Uses ``starlette.testclient.TestClient`` because it boots the ASGI app
+end-to-end and handles the lifespan protocol automatically — ``httpx``
+alone does not run lifespan events, which would mask the exact kind of
+misconfiguration this test suite exists to catch.
+"""
+
+import pytest
+
+
+@pytest.mark.unit
+def test_asgi_app_imports():
+    from application.asgi import asgi_app
+
+    assert asgi_app is not None
+
+
+@pytest.mark.unit
+def test_flask_route_served_through_starlette_mount():
+    """GET /api/health should reach the Flask app via a2wsgi and return 200."""
+    from starlette.testclient import TestClient
+
+    from application.asgi import asgi_app
+
+    with TestClient(asgi_app) as client:
+        r = client.get("/api/health")
+    assert r.status_code == 200
+    assert r.json() == {"status": "ok"}
+
+
+@pytest.mark.unit
+def test_mcp_endpoint_mounted_and_lifespan_runs():
+    """/mcp must be reachable AND the FastMCP session manager must start.
+
+    Without ``lifespan=mcp_app.lifespan`` on the outer Starlette app,
+    every /mcp request raises ``RuntimeError: Task group is not
+    initialized``. Hitting the endpoint under a real lifespan-aware
+    client catches that.
+    """
+    from starlette.testclient import TestClient
+
+    from application.asgi import asgi_app
+
+    with TestClient(asgi_app) as client:
+        # Minimal MCP initialize request. Doesn't need to succeed — we
+        # just need a non-404, non-500-with-RuntimeError response to
+        # confirm the mount + lifespan are both wired.
+        r = client.post(
+            "/mcp/",
+            headers={
+                "Origin": "http://example.com",
+                "Content-Type": "application/json",
+                "Accept": "application/json, text/event-stream",
+            },
+            json={
+                "jsonrpc": "2.0",
+                "id": 1,
+                "method": "initialize",
+                "params": {
+                    "protocolVersion": "2025-03-26",
+                    "capabilities": {},
+                    "clientInfo": {"name": "pytest", "version": "0"},
+                },
+            },
+        )
+    assert r.status_code != 404, f"/mcp mount unreachable: {r.status_code}"
+    # A successful initialize returns 200 with a Mcp-Session-Id header.
+    assert r.status_code == 200
+    assert "mcp-session-id" in {k.lower() for k in r.headers.keys()}
+    assert r.headers.get("access-control-expose-headers") == "Mcp-Session-Id"
+
+
+@pytest.mark.unit
+def test_cors_headers_on_flask_route():
+    """CORS middleware should emit allow-origin on actual (non-preflight) requests.
+
+    ``allow_origins=["*"]`` → header value is literal ``*`` (not an echo).
+    """
+    from starlette.testclient import TestClient
+
+    from application.asgi import asgi_app
+
+    with TestClient(asgi_app) as client:
+        r = client.get("/api/health", headers={"Origin": "http://example.com"})
+    assert r.status_code == 200
+    assert r.headers.get("access-control-allow-origin") == "*"
+
+
+@pytest.mark.unit
+def test_cors_preflight_on_flask_route():
+    """OPTIONS preflight on a Flask route should be handled by Starlette CORSMiddleware."""
+    from starlette.testclient import TestClient
+
+    from application.asgi import asgi_app
+
+    with TestClient(asgi_app) as client:
+        r = client.options(
+            "/api/health",
+            headers={
+                "Origin": "http://example.com",
+                "Access-Control-Request-Method": "GET",
+                "Access-Control-Request-Headers": "Content-Type",
+            },
+        )
+    assert r.status_code in (200, 204)
+    assert r.headers.get("access-control-allow-origin") == "*"
+    assert "GET" in r.headers.get("access-control-allow-methods", "")
+
+
+@pytest.mark.unit
+def test_cors_preflight_on_mcp_route():
+    """Browser clients hitting /mcp should be allowed to send session headers."""
+    from starlette.testclient import TestClient
+
+    from application.asgi import asgi_app
+
+    with TestClient(asgi_app) as client:
+        r = client.options(
+            "/mcp/",
+            headers={
+                "Origin": "http://example.com",
+                "Access-Control-Request-Method": "POST",
+                "Access-Control-Request-Headers": (
+                    "Authorization, Content-Type, Mcp-Session-Id"
+                ),
+            },
+        )
+    assert r.status_code in (200, 204)
+    assert r.headers.get("access-control-allow-origin") == "*"
+    assert "Mcp-Session-Id" in r.headers.get("access-control-allow-headers", "")