Modal MCP Server
모달 MCP 서버
Cursor 내에서 Modal 볼륨과 상호 작용하고 Modal 애플리케이션을 배포하기 위한 MCP 서버 구현입니다.
설치
이 저장소를 복제하세요:
지엑스피1
uv사용하여 종속성을 설치합니다.
uv syncRelated MCP server: Outsource MCP
구성
Cursor에서 이 MCP 서버를 사용하려면 ~/.cursor/mcp.json 에 다음 구성을 추가하세요.
{
"mcpServers": {
"modal-mcp-server": {
"command": "uv",
"args": [
"--project", "/path/to/modal-mcp-server",
"run", "/path/to/modal-mcp-server/src/modal_mcp/server.py"
]
}
}
}/path/to/modal-mcp-server 복제된 저장소의 절대 경로로 바꾸세요.
요구 사항
Python 3.11 이상
uv패키지 관리자유효한 자격 증명으로 구성된 모달 CLI
Modal 배포 지원:
배포되는 프로젝트는 종속성 관리를 위해
uv사용해야 합니다.Modal은 프로젝트의 가상 환경에 설치되어야 합니다.
지원되는 도구
모달 볼륨 작업
모달 볼륨 목록 (
list_modal_volumes)사용자 환경의 모든 모달 볼륨을 나열합니다.
JSON 형식의 볼륨 정보를 반환합니다.
매개변수: 없음
볼륨 내용 목록 (
list_modal_volume_contents)모달 볼륨의 파일 및 디렉토리를 나열합니다.
매개변수:
volume_name: 모달 볼륨의 이름path: 볼륨 내 경로(기본값: "/")
파일 복사 (
copy_modal_volume_files)모달 볼륨 내의 파일을 복사합니다.
매개변수:
volume_name: 모달 볼륨의 이름paths: 마지막 경로가 목적지인 경로 목록
예:
["source.txt", "dest.txt"]또는["file1.txt", "file2.txt", "dest_dir/"]
파일 제거 (
remove_modal_volume_file)모달 볼륨에서 파일이나 디렉토리를 삭제합니다.
매개변수:
volume_name: 모달 볼륨의 이름remote_path: 삭제할 파일/디렉토리 경로recursive: 재귀적 삭제에 대한 부울 플래그(기본값: false)
파일 업로드 (
put_modal_volume_file)모달 볼륨에 파일이나 디렉토리를 업로드합니다.
매개변수:
volume_name: 모달 볼륨의 이름local_path: 업로드할 로컬 파일/디렉토리 경로remote_path: 업로드할 볼륨의 경로(기본값: "/")force: 기존 파일을 덮어쓸지 여부를 지정하는 부울 플래그(기본값: false)
파일 다운로드 (
get_modal_volume_file)모달 볼륨에서 파일을 다운로드합니다.
매개변수:
volume_name: 모달 볼륨의 이름remote_path: 다운로드할 볼륨의 파일/디렉토리 경로local_destination: 다운로드한 파일을 저장할 로컬 경로(기본값: 현재 디렉토리)force: 기존 파일을 덮어쓸지 여부를 지정하는 부울 플래그(기본값: false)
참고: 파일 내용을 stdout에 쓰려면
local_destination으로 "-"를 사용하세요.
모달 배포
모달 앱 배포 (
deploy_modal_app)모달 애플리케이션을 배포합니다
매개변수:
absolute_path_to_app: Modal 애플리케이션 파일에 대한 절대 경로
참고: Modal 앱이 포함된 프로젝트는 다음과 같아야 합니다.
종속성 관리를 위해
uv사용하세요가상 환경에
modalCLI를 설치하세요
응답 형식
모든 도구는 작업 유형에 따라 약간의 차이가 있지만 표준화된 형식으로 응답을 반환합니다.
# JSON operations (list volumes, list contents):
{
"success": True,
"data": {...} # JSON data from Modal CLI
}
# File operations (put, get, copy, remove):
{
"success": True,
"message": "Operation successful message",
"command": "executed command string",
"stdout": "command output", # if any
"stderr": "error output" # if any
}
# Error case (all operations):
{
"success": False,
"error": "Error message describing what went wrong",
"command": "executed command string", # for file operations
"stdout": "command output", # if available
"stderr": "error output" # if available
}기여하다
기여를 환영합니다! 풀 리퀘스트를 제출해 주세요.
특허
이 프로젝트는 MIT 라이선스에 따라 라이선스가 부여되었습니다. 자세한 내용은 라이선스 파일을 참조하세요.
Available Tools
7 toolscopy_modal_volume_filesA
Copy files within a Modal volume. Can copy a source file to a destination file
or multiple source files to a destination directory.
Args:
volume_name: Name of the Modal volume to perform copy operation in.
paths: List of paths for the copy operation. The last path is the destination,
all others are sources. For example: ["source1.txt", "source2.txt", "dest_dir/"]
Returns:
A dictionary containing the result of the copy operation.
Raises:
Exception: If the copy operation fails for any reason.
Inputs: volume_name (required), paths (required).
deploy_modal_appC
Deploy a Modal application using the provided parameters.
Args:
absolute_path_to_app: The absolute path to the Modal application to deploy.
Returns:
A dictionary containing deployment results.
Raises:
Exception: If deployment fails for any reason.
Inputs: absolute_path_to_app (required).
get_modal_volume_fileA
Download files from a Modal volume.
Args:
volume_name: Name of the Modal volume to download from.
remote_path: Path to the file or directory in the volume to download.
local_destination: Local path to save the downloaded file(s). Defaults to current directory.
Use "-" to write file contents to stdout.
force: If True, overwrite existing files. Defaults to False.
Returns:
A dictionary containing the result of the download operation.
Raises:
Exception: If the download operation fails for any reason.
Inputs: volume_name (required), remote_path (required), local_destination, force.
list_modal_volume_contentsB
List files and directories in a Modal volume.
Args:
volume_name: Name of the Modal volume to list contents from.
path: Path within the volume to list contents from. Defaults to root ("/").
Returns:
A dictionary containing the parsed JSON output of the volume contents.
Inputs: volume_name (required), path.
list_modal_volumesB
List all Modal volumes using the Modal CLI with JSON output.
Returns:
A dictionary containing the parsed JSON output of the Modal volumes list.
Inputs: no named parameters.
put_modal_volume_fileA
Upload a file or directory to a Modal volume.
Args:
volume_name: Name of the Modal volume to upload to.
local_path: Path to the local file or directory to upload.
remote_path: Path in the volume to upload to. Defaults to root ("/").
If ending with "/", it's treated as a directory and the file keeps its name.
force: If True, overwrite existing files. Defaults to False.
Returns:
A dictionary containing the result of the upload operation.
Raises:
Exception: If the upload operation fails for any reason.
Inputs: volume_name (required), local_path (required), remote_path, force.
remove_modal_volume_fileB
Delete a file or directory from a Modal volume.
Args:
volume_name: Name of the Modal volume to delete from.
remote_path: Path to the file or directory to delete.
recursive: If True, delete directories recursively. Required for deleting directories.
Returns:
A dictionary containing the result of the delete operation.
Raises:
Exception: If the delete operation fails for any reason.
Inputs: volume_name (required), remote_path (required), recursive.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 7 tools
Every tool has a clearly distinct purpose with no overlap: copy, deploy, get (download), list contents, list volumes, put (upload), and remove (delete) are all unique operations. The descriptions clearly differentiate between volume operations and app deployment, making tool selection straightforward for an agent.
All tools follow a consistent verb_noun pattern with 'modal' as a prefix (e.g., copy_modal_volume_files, deploy_modal_app, get_modal_volume_file). The naming is uniform throughout, using snake_case and clear action-object pairs, making the set predictable and easy to understand.
With 7 tools, this server is well-scoped for managing Modal volumes and applications. Each tool serves a specific, necessary function (e.g., CRUD operations for volumes, listing, and deployment), and there are no redundant or missing tools for the apparent domain, making the count ideal.
The tool surface provides complete coverage for Modal volume management (create via put, read via get and list, update via copy, delete via remove) and app deployment. There are no obvious gaps; agents can perform all core workflows without dead ends, ensuring effective operation within the domain.
Maintenance
Related MCP Connectors
MCP server for AI agents to plan, verify, and deploy Cloudflare-native apps.
MCP server for building and testing AI agents with multi-model experimentation and insights.
MCP server that lets your AI assistant build and deploy apps, hosting each one at a live URL.
MCP server for your apps' tools and custom tools, plus hosted AI agents and approval-gated workflows
Related MCP Servers
- FlicenseDqualityFmaintenanceAn MCP server that allows users to deploy Python scripts to Modal directly from Claude, providing a link to the deployed application that can be shared with others.14-
- AlicenseAqualityDmaintenanceAn MCP server that enables AI applications to access 20+ model providers (including OpenAI, Anthropic, Google) through a unified interface for text and image generation.230MIT
- AlicenseAqualityBmaintenanceAn MCP server for managing Modal apps, containers, volumes, and secrets. Also used for deploying & running Modal apps directly from Claude Code and other MCP clients.12396 PyPI4MIT
- AlicenseBqualityDmaintenanceAn MCP server that enables LLM coding agents to deploy web applications to AWS serverless infrastructure, supporting backend, frontend, and fullstack deployments.51MIT