feat: Split MCP into Query and File servers, add tests, configure Tesseract French (BRIC-10)

.gitignore

@@ -15,4 +15,5 @@ trivy-report-fixed.json
 coverage.xml
 .ruff_cache
 .pytest_cache
-trivy-report-current.json
+trivy-report-current.json
+.vscode

Dockerfile

@@ -22,6 +22,7 @@ RUN apt-get update && apt-get install -y \
     libgl1 \
     git \
     tesseract-ocr \
+    tesseract-ocr-fra \
     && rm -rf /var/lib/apt/lists/*
 
 # Set working directory

README.md

@@ -14,23 +14,24 @@ Multi-modal RAG service exposing a REST API and MCP server for document indexing
                             |
             +---------------+---------------+
             |                               |
-   Application Layer                  MCP Tools
-   +------------------------------+   (FastMCP)
-   | api/                         |       |
-   |   indexing_routes.py         |       |
-   |   query_routes.py           |       |
-   |   file_routes.py             |       |
-   |   health_routes.py          |       |
-   | use_cases/                   |       |
-   |   IndexFileUseCase           |       |
-   |   IndexFolderUseCase         |       |
-   |   QueryUseCase               |       |
-   |   ListFilesUseCase            |       |
-   |   ReadFileUseCase             |       |
-   | requests/ responses/         |       |
-   +------------------------------+       |
-            |         |          |        |
-            v         v          v        v
+    Application Layer            MCP Servers (FastMCP)
+    +------------------------------+       |
+    | api/                         |   +---+--------+  +--+-----------+
+    |   indexing_routes.py         |   | RAGAnything |  | RAGAnything |
+    |   query_routes.py            |   | Query       |  | Files       |
+    |   file_routes.py             |   |  /rag/mcp   |  |  /files/mcp |
+    |   health_routes.py           |   +---+--------+  +--+-----------+
+    | use_cases/                   |       |               |
+    |   IndexFileUseCase           |   query_knowledge    list_files
+    |   IndexFolderUseCase         |   _base              read_file
+    |   QueryUseCase               |   query_knowledge
+    |   ListFilesUseCase           |   _base_multimodal
+    |   ListFoldersUseCase         |
+    |   ReadFileUseCase            |
+    | requests/ responses/         |
+    +------------------------------+
+             |         |          |
+             v         v          v
    Domain Layer (ports)
    +------------------------------------------+
    | RAGEnginePort  StoragePort  BM25EnginePort  DocumentReaderPort |
@@ -173,7 +174,7 @@ The service automatically detects and processes the following document formats t
 
 | Format | Extensions | Notes |
 |--------|------------|-------|
-| PDF | `.pdf` | Includes OCR support |
+| PDF | `.pdf` | Includes OCR support (English + French via Tesseract) |
 | Microsoft Word | `.docx` | |
 | Microsoft PowerPoint | `.pptx` | |
 | Microsoft Excel | `.xlsx` | |
@@ -213,6 +214,26 @@ Response (`200 OK`):
 | `prefix` | string | `""` | MinIO prefix to filter files by |
 | `recursive` | boolean | `true` | List files in subdirectories |
 
+### List folders
+
+Returns top-level folder prefixes in the bucket. REST-only endpoint (not exposed as an MCP tool).
+
+```bash
+curl http://localhost:8000/api/v1/files/folders
+```
+
+Response (`200 OK`):
+
+```json
+["documents/", "photos/", "reports/"]
+```
+
+Error responses:
+
+| Status | Condition |
+|--------|-----------|
+| `404` | Bucket not found |
+
 ### Read a file
 
 Downloads the file from MinIO, extracts its text content using Kreuzberg, and returns the result. Supports 91 file formats including PDF, Office documents, images, and HTML.
@@ -382,20 +403,24 @@ Response (`200 OK`):
 
 The `combined_score` is the sum of `bm25_score` and `vector_score`, each computed as `1 / (k + rank)`. Results are sorted by `combined_score` descending. A chunk that appears in both result sets will have a higher combined score than one that appears in only one.
 
-## MCP Server
+## MCP Servers
 
-The MCP server is mounted at `/mcp` and exposes the following tools:
+The service exposes **two separate MCP servers**, both using streamable HTTP transport:
 
-### Tool: `query_knowledge_base`
+### RAGAnythingQuery — `/rag/mcp`
+
+Query-focused tools for searching the indexed knowledge base.
+
+#### Tool: `query_knowledge_base`
 
 | Parameter | Type | Default | Description |
 |-----------|------|---------|-------------|
 | `working_dir` | string | required | RAG workspace directory for this project |
 | `query` | string | required | The search query |
-| `mode` | string | `"naive"` | Search mode: `naive`, `local`, `global`, `hybrid`, `hybrid+`, `mix`, `bm25`, `bypass` |
-| `top_k` | integer | `10` | Number of chunks to retrieve |
+| `mode` | string | `"hybrid"` | Search mode: `naive`, `local`, `global`, `hybrid`, `hybrid+`, `mix`, `bm25`, `bypass` |
+| `top_k` | integer | `5` | Number of chunks to retrieve |
 
-### Tool: `query_knowledge_base_multimodal`
+#### Tool: `query_knowledge_base_multimodal`
 
 | Parameter | Type | Default | Description |
 |-----------|------|---------|-------------|
@@ -405,54 +430,32 @@ The MCP server is mounted at `/mcp` and exposes the following tools:
 | `mode` | string | `"hybrid"` | Search mode |
 | `top_k` | integer | `5` | Number of chunks to retrieve |
 
-### Tool: `list_files`
+### RAGAnythingFiles — `/files/mcp`
+
+File browsing tools for listing and reading files from MinIO storage.
+
+#### Tool: `list_files`
 
 | Parameter | Type | Default | Description |
 |-----------|------|---------|-------------|
 | `prefix` | string | `""` | MinIO prefix to filter files by |
 | `recursive` | boolean | `true` | List files in subdirectories |
 
-### Tool: `read_file`
+#### Tool: `read_file`
 
 | Parameter | Type | Default | Description |
 |-----------|------|---------|-------------|
 | `file_path` | string | required | File path in MinIO bucket (e.g. `documents/report.pdf`) |
 
 Downloads the file from MinIO, extracts its text content using Kreuzberg, and returns the extracted text along with metadata and any detected tables.
 
-### Transport modes
-
-The `MCP_TRANSPORT` environment variable controls how the MCP server is exposed:
-
-| Value | Behavior |
-|-------|----------|
-| `stdio` | MCP runs over stdin/stdout; FastAPI runs in a background thread |
-| `sse` | MCP mounted at `/mcp` as SSE endpoint |
-| `streamable` | MCP mounted at `/mcp` as streamable HTTP endpoint |
+### Transport
 
-### Claude Desktop configuration
+Both MCP servers use **streamable HTTP** transport exclusively. Connect MCP clients to the mount paths:
 
-Add to `~/Library/Application Support/Claude/claude_desktop_config.json`:
-
-```json
-{
-  "mcpServers": {
-    "raganything": {
-      "command": "uv",
-      "args": [
-        "run",
-        "--directory",
-        "/absolute/path/to/mcp-raganything",
-        "python",
-        "-m",
-        "src.main"
-      ],
-      "env": {
-        "MCP_TRANSPORT": "stdio"
-      }
-    }
-  }
-}
+```
+http://localhost:8000/rag/mcp      # RAGAnythingQuery
+http://localhost:8000/files/mcp    # RAGAnythingFiles
 ```
 
 ## Configuration
@@ -465,7 +468,6 @@ All configuration is via environment variables, loaded through Pydantic Settings
 |----------|---------|-------------|
 | `HOST` | `0.0.0.0` | Server bind address |
 | `PORT` | `8000` | Server port |
-| `MCP_TRANSPORT` | `stdio` | MCP transport: `stdio`, `sse`, `streamable` |
 | `ALLOWED_ORIGINS` | `["*"]` | CORS allowed origins |
 | `OUTPUT_DIR` | system temp | Temporary directory for downloaded files |
 | `UVICORN_LOG_LEVEL` | `critical` | Uvicorn log level |
@@ -577,7 +579,7 @@ The PostgreSQL server must have the `pg_textsearch` extension installed and load
 
 ```
 src/
-  main.py                           -- FastAPI app, MCP mount, entry point
+  main.py                           -- FastAPI app, dual MCP mounts, entry point
   config.py                         -- Pydantic Settings config classes
   dependencies.py                   -- Dependency injection wiring
   domain/
@@ -593,8 +595,9 @@ src/
       health_routes.py               -- GET /health
       indexing_routes.py              -- POST /file/index, /folder/index
       query_routes.py                 -- POST /query
-      file_routes.py                  -- GET /files/list, POST /files/read
-      mcp_tools.py                    -- MCP tools: query_knowledge_base, list_files, read_file
+      file_routes.py                  -- GET /files/list, GET /files/folders, POST /files/read
+      mcp_query_tools.py              -- MCP tools: query_knowledge_base, query_knowledge_base_multimodal
+      mcp_file_tools.py                -- MCP tools: list_files, read_file
     requests/
       indexing_request.py            -- IndexFileRequest, IndexFolderRequest
       query_request.py                -- QueryRequest, MultimodalQueryRequest
@@ -607,6 +610,7 @@ src/
       index_folder_use_case.py       -- Downloads from MinIO, indexes folder
       query_use_case.py              -- Query with bm25/hybrid+ support
       list_files_use_case.py          -- Lists files with metadata from MinIO
+      list_folders_use_case.py        -- Lists folder prefixes from MinIO
       read_file_use_case.py           -- Reads file from MinIO, extracts content via Kreuzberg
   infrastructure/
     rag/

pyproject.toml

@@ -14,7 +14,7 @@ dependencies = [
     "fastmcp>=3.2.0",
     "cryptography>=46.0.5",
     "httpx>=0.27.0",
-    "kreuzberg>=4.0.0",
+    "kreuzberg[all]>=4.8.2",
     "lightrag-hku>=1.4.13",
     "lightrag-hku[api]>=1.4.13",
     "mcp>=1.24.0",

src/application/api/file_routes.py

@@ -5,15 +5,19 @@
 from application.requests.file_request import ReadFileRequest
 from application.responses.file_response import FileContentResponse, FileInfoResponse
 from application.use_cases.list_files_use_case import ListFilesUseCase
+from application.use_cases.list_folders_use_case import ListFoldersUseCase
 from application.use_cases.read_file_use_case import ReadFileUseCase
-from dependencies import get_list_files_use_case, get_read_file_use_case
+from dependencies import (
+    get_list_files_use_case,
+    get_list_folders_use_case,
+    get_read_file_use_case,
+)
 
 file_router = APIRouter(tags=["Files"])
 
 
 @file_router.get(
     "/files/list",
-    response_model=list[FileInfoResponse],
     status_code=status.HTTP_200_OK,
 )
 async def list_files(
@@ -25,9 +29,24 @@ async def list_files(
     return [FileInfoResponse(**asdict(f)) for f in files]
 
 
+@file_router.get(
+    "/files/folders",
+    status_code=status.HTTP_200_OK,
+)
+async def list_folders(
+    use_case: ListFoldersUseCase = Depends(get_list_folders_use_case),
+) -> list[str]:
+    try:
+        return await use_case.execute()
+    except FileNotFoundError as e:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=str(e),
+        ) from None
+
+
 @file_router.post(
     "/files/read",
-    response_model=FileContentResponse,
     status_code=status.HTTP_200_OK,
 )
 async def read_file(

src/application/api/health_routes.py

@@ -1,4 +1,8 @@
-from fastapi import APIRouter
+from fastapi import APIRouter, Depends
+from fastapi.responses import JSONResponse
+
+from application.use_cases.liveness_check_use_case import LivenessCheckUseCase
+from dependencies import get_liveness_check_use_case
 
 health_router = APIRouter(tags=["Health"])
 
@@ -12,3 +16,17 @@ def health_check():
         dict: Status message indicating the API is running.
     """
     return {"message": "RAG Anything API is running"}
+
+
+@health_router.get("/health/live")
+async def liveness_check(
+    use_case: LivenessCheckUseCase = Depends(get_liveness_check_use_case),
+):
+    """Liveness probe that checks PostgreSQL and MinIO connectivity.
+
+    Returns:
+        200 if both connections are healthy, 503 if any is unreachable.
+    """
+    result = await use_case.execute()
+    status_code = 200 if result["status"] == "healthy" else 503
+    return JSONResponse(content=result, status_code=status_code)

src/application/api/mcp_file_tools.py

@@ -0,0 +1,65 @@
+"""MCP file tools for RAGAnything.
+
+These tools are registered with FastMCP for Claude Desktop integration.
+"""
+
+import logging
+from dataclasses import asdict
+
+from fastmcp import FastMCP
+
+from application.responses.file_response import FileContentResponse, FileInfoResponse
+from dependencies import (
+    get_list_files_use_case,
+    get_read_file_use_case,
+)
+
+logger = logging.getLogger(__name__)
+
+mcp_files = FastMCP("RAGAnythingFiles")
+
+
+@mcp_files.tool()
+async def list_files(
+    prefix: str = "", recursive: bool = True
+) -> list[FileInfoResponse]:
+    """List files in MinIO storage under a given prefix.
+
+    Args:
+        prefix: MinIO prefix/path to filter files by (e.g. 'documents/')
+        recursive: Whether to list files in subdirectories (default True)
+
+    Returns:
+        List of file objects with object_name, size, and last_modified
+    """
+    use_case = get_list_files_use_case()
+    files = await use_case.execute(prefix=prefix, recursive=recursive)
+    return [FileInfoResponse(**asdict(f)) for f in files]
+
+
+@mcp_files.tool()
+async def read_file(file_path: str) -> FileContentResponse:
+    """Read and extract text content from a file stored in MinIO.
+
+    Supports 91 file formats including PDF, Office documents, images, HTML, etc.
+    Uses Kreuzberg for document intelligence extraction.
+
+    Args:
+        file_path: Path to the file in MinIO bucket (e.g. 'documents/report.pdf')
+
+    Returns:
+        Extracted text content with metadata and any detected tables
+    """
+    use_case = get_read_file_use_case()
+    try:
+        result = await use_case.execute(file_path=file_path)
+    except FileNotFoundError:
+        raise ValueError(f"File not found: {file_path}") from None
+    except Exception:
+        logger.exception("Unexpected error reading file: %s", file_path)
+        raise RuntimeError("Failed to read file") from None
+    return FileContentResponse(
+        content=result.content,
+        metadata=result.metadata,
+        tables=result.tables,
+    )