feat(cli): Add files_ls and files_rm tools to MCP server (#235)

* feat(cli): Add `files_ls` and `files_rm` tools to MCP server

* docs(cli): Document `files_ls` and `files_rm` tools in cli.md

* Auto generate docs

---------

Co-authored-by: Davidyz <Davidyz@users.noreply.github.com>

Author: Davidyz

doc/VectorCode-cli.txt

@@ -803,7 +803,9 @@ features:
 - `ls`list local collections, similar to the `ls` subcommand in the CLI;
 - `query`query from a given collection, similar to the `query` subcommand in
     the CLI;
-- `vectorise`vectorise files into a given project.
+- `vectorise`vectorise files into a given project;
+- `files_ls`show files that have been indexed for the current project;
+- `files_rm`remove some files from the database for a project.
 
 To try it out, install the `vectorcode[mcp]` dependency group and the MCP
 server is available in the shell as `vectorcode-mcp-server`.

docs/cli.md

@@ -726,7 +726,9 @@ features:
 - `ls`: list local collections, similar to the `ls` subcommand in the CLI;
 - `query`: query from a given collection, similar to the `query` subcommand in
   the CLI;
-- `vectorise`: vectorise files into a given project.
+- `vectorise`: vectorise files into a given project;
+- `files_ls`: show files that have been indexed for the current project;
+- `files_rm`: remove some files from the database for a project.
 
 To try it out, install the `vectorcode[mcp]` dependency group and the MCP server 
 is available in the shell as `vectorcode-mcp-server`. 

src/vectorcode/mcp_main.py

@@ -5,9 +5,10 @@
 import sys
 from dataclasses import dataclass
 from pathlib import Path
-from typing import Optional
+from typing import Optional, cast
 
 import shtab
+from chromadb.types import Where
 
 from vectorcode.subcommands.vectorise import (
     VectoriseStats,
@@ -33,11 +34,17 @@
     cleanup_path,
     config_logging,
     expand_globs,
+    expand_path,
     find_project_config_dir,
     get_project_config,
     load_config_file,
 )
-from vectorcode.common import ClientManager, get_collection, get_collections
+from vectorcode.common import (
+    ClientManager,
+    get_collection,
+    get_collections,
+    list_collection_files,
+)
 from vectorcode.subcommands.prompt import prompt_by_categories
 from vectorcode.subcommands.query import get_query_result_files
 
@@ -224,6 +231,34 @@ async def query_tool(
         )
 
 
+async def ls_files(project_root: str) -> list[str]:
+    """
+    project_root: Directory to the repository. MUST be from the vectorcode `ls` tool or user input;
+    """
+    configs = await get_project_config(expand_path(project_root, True))
+    async with ClientManager().get_client(configs) as client:
+        return await list_collection_files(await get_collection(client, configs, False))
+
+
+async def rm_files(files: list[str], project_root: str):
+    """
+    files: list of paths of the files to be removed;
+    project_root: Directory to the repository. MUST be from the vectorcode `ls` tool or user input;
+    """
+    configs = await get_project_config(expand_path(project_root, True))
+    async with ClientManager().get_client(configs) as client:
+        try:
+            collection = await get_collection(client, configs, False)
+            files = [str(expand_path(i, True)) for i in files if os.path.isfile(i)]
+            if files:
+                await collection.delete(where=cast(Where, {"path": {"$in": files}}))
+            else:  # pragma: nocover
+                logger.warning(f"All paths were invalid: {files}")
+        except ValueError:  # pragma: nocover
+            logger.warning(f"Failed to find the collection at {configs.project_root}")
+            return
+
+
 async def mcp_server():
     global default_config, default_project_root
 
@@ -283,6 +318,18 @@ async def mcp_server():
         ),
     )
 
+    mcp.add_tool(
+        fn=rm_files,
+        name="files_rm",
+        description="Remove files from VectorCode embedding database.",
+    )
+
+    mcp.add_tool(
+        fn=ls_files,
+        name="files_ls",
+        description="List files that have been indexed by VectorCode.",
+    )
+
     return mcp
 
 

tests/test_mcp.py

@@ -7,12 +7,15 @@
 from mcp import McpError
 
 from vectorcode.cli_utils import Config
+from vectorcode.common import ClientManager
 from vectorcode.mcp_main import (
     get_arg_parser,
     list_collections,
+    ls_files,
     mcp_server,
     parse_cli_args,
     query_tool,
+    rm_files,
     vectorise_files,
 )
 
@@ -335,7 +338,7 @@ async def test_mcp_server():
 
         await mcp_server()
 
-        assert mock_add_tool.call_count == 3
+        assert mock_add_tool.call_count == 5
 
 
 @pytest.mark.asyncio
@@ -374,10 +377,68 @@ async def new_get_collections(clients):
 
         await mcp_server()
 
-        assert mock_add_tool.call_count == 3
+        assert mock_add_tool.call_count == 5
         mock_get_collections.assert_called()
 
 
+@pytest.mark.asyncio
+async def test_ls_files_success():
+    ClientManager().clear()
+    mock_client = MagicMock()
+    mock_collection = MagicMock()
+    expected_files = ["/test/project/file1.py", "/test/project/dir/file2.txt"]
+
+    with (
+        patch("vectorcode.mcp_main.get_project_config") as mock_get_project_config,
+        patch(
+            "vectorcode.mcp_main.ClientManager._create_client", return_value=mock_client
+        ),
+        patch("vectorcode.common.try_server", return_value=True),
+        patch("vectorcode.mcp_main.get_collection", return_value=mock_collection),
+        patch(
+            "vectorcode.mcp_main.list_collection_files", return_value=expected_files
+        ) as mock_list_collection_files,
+        patch(
+            "vectorcode.cli_utils.expand_path", side_effect=lambda x, y: x
+        ),  # Mock expand_path to return input
+    ):
+        mock_get_project_config.return_value = Config(project_root="/test/project")
+        result = await ls_files(project_root="/test/project")
+
+        assert result == expected_files
+        mock_get_project_config.assert_called_once_with("/test/project")
+
+        mock_list_collection_files.assert_called_once_with(mock_collection)
+
+
+@pytest.mark.asyncio
+async def test_rm_files_success():
+    ClientManager().clear()
+    mock_client = MagicMock()
+    mock_collection = MagicMock()
+    files_to_remove = ["/test/project/file1.py", "/test/project/file2.txt"]
+
+    with (
+        patch("os.path.isfile", side_effect=lambda x: x in files_to_remove),
+        patch("vectorcode.mcp_main.get_project_config") as mock_get_project_config,
+        patch(
+            "vectorcode.mcp_main.ClientManager._create_client", return_value=mock_client
+        ),
+        patch("vectorcode.common.try_server", return_value=True),
+        patch("vectorcode.mcp_main.get_collection", return_value=mock_collection),
+        patch("vectorcode.cli_utils.expand_path", side_effect=lambda x, y: x),
+    ):
+        mock_get_project_config.return_value = Config(project_root="/test/project")
+        mock_collection.delete = AsyncMock()
+
+        await rm_files(files=files_to_remove, project_root="/test/project")
+
+        mock_get_project_config.assert_called_once_with("/test/project")
+        mock_collection.delete.assert_called_once_with(
+            where={"path": {"$in": files_to_remove}}
+        )
+
+
 def test_arg_parser():
     assert isinstance(get_arg_parser(), ArgumentParser)