GDP MCP Tools
MCP server for Grubhub Data Platform operations, providing 40+ tools across 18 service categories for data platform management, observability, analytics, and collaboration.
README
GDP MCP Tools
MCP (Model Context Protocol) server for Grubhub Data Platform operations. 40+ tools across 18 service categories. Works with Claude Code, VS Code, Cursor, JetBrains, or any MCP-compatible client.
Quick Start (5 minutes)
1. Clone and set up virtual environment
# Ask your team lead for the repo URL, or use:
git clone https://github.com/asuar16/gdp-mcp-tools.git
cd gdp-mcp-tools
# macOS / Linux
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
# Windows (PowerShell)
python -m venv venv
.\venv\Scripts\Activate.ps1
pip install -r requirements.txt
2. Configure credentials
cp .env.example src/.env
Edit src/.env with your credentials. See Credentials below.
3. Configure your MCP client
Add to .mcp.json in any repo where you want these tools:
macOS / Linux:
{
"mcpServers": {
"gdp-tools": {
"type": "stdio",
"command": "/absolute/path/to/gdp-mcp-tools/venv/bin/python",
"args": ["/absolute/path/to/gdp-mcp-tools/src/server.py"]
}
}
}
Windows:
{
"mcpServers": {
"gdp-tools": {
"type": "stdio",
"command": "C:\\path\\to\\gdp-mcp-tools\\venv\\Scripts\\python.exe",
"args": ["C:\\path\\to\\gdp-mcp-tools\\src\\server.py"]
}
}
}
4. Connect VPN and start using
All tools require VPN. Connect first, then use any MCP client.
Using from Any Repo
Already have gdp-mcp-tools set up? Here's how to add it to a new repo in 1 minute:
Step 1: Get your paths
cd ~/path/to/gdp-mcp-tools
echo "Python: $(pwd)/venv/bin/python"
echo "Server: $(pwd)/src/server.py"
Step 2: Create .mcp.json in your new repo
cd ~/path/to/your-new-repo
cat > .mcp.json << 'EOF'
{
"mcpServers": {
"gdp-tools": {
"type": "stdio",
"command": "/absolute/path/to/gdp-mcp-tools/venv/bin/python",
"args": ["/absolute/path/to/gdp-mcp-tools/src/server.py"]
}
}
}
EOF
Replace the paths with your actual paths from Step 1.
Step 3: Start your AI coding session
Claude Code:
cd ~/path/to/your-new-repo
claude
# GDP tools will auto-connect. Try: "check vpn status"
VS Code / Cursor:
Open the repo folder -> MCP tools appear in Copilot/AI sidebar
Gemini CLI:
cd ~/path/to/your-new-repo
gemini
# Tools auto-load from .gemini/settings.json
Antigravity:
Open project -> Tools auto-load from mcp_config.json
That's it. The .mcp.json tells any MCP client where to find the server. No installation needed in the new repo -- everything runs from gdp-mcp-tools/.
Example: Adding to events-mart, ods_data, or any repo
{
"mcpServers": {
"gdp-tools": {
"type": "stdio",
"command": "/Users/yourname/gdp-mcp-tools/venv/bin/python",
"args": ["/Users/yourname/gdp-mcp-tools/src/server.py"]
}
}
}
Works from any repo -- events-mart, ods_data, or even non-GDP projects.
Client Setup
Claude Code (CLI / Desktop / Web)
Add .mcp.json to your project root:
{
"mcpServers": {
"gdp-tools": {
"type": "stdio",
"command": "/absolute/path/to/gdp-mcp-tools/venv/bin/python",
"args": ["/absolute/path/to/gdp-mcp-tools/src/server.py"]
}
}
}
VS Code / Cursor
Add to .vscode/settings.json:
{
"mcp.servers": {
"gdp-tools": {
"type": "stdio",
"command": "/absolute/path/to/gdp-mcp-tools/venv/bin/python",
"args": ["/absolute/path/to/gdp-mcp-tools/src/server.py"]
}
}
}
Google Antigravity
Add to ~/.gemini/antigravity/mcp_config.json (or project-level .gemini/settings.json):
macOS / Linux:
{
"mcpServers": {
"gdp-tools": {
"command": "/absolute/path/to/gdp-mcp-tools/venv/bin/python",
"args": ["/absolute/path/to/gdp-mcp-tools/src/server.py"]
}
}
}
Windows:
{
"mcpServers": {
"gdp-tools": {
"command": "C:\\path\\to\\gdp-mcp-tools\\venv\\Scripts\\python.exe",
"args": ["C:\\path\\to\\gdp-mcp-tools\\src\\server.py"]
}
}
}
To reload after config change: use Manage MCP Servers > View raw config in Antigravity UI, or restart.
Gemini CLI
Add to ~/.gemini/settings.json (global) or .gemini/settings.json (project-level):
macOS / Linux:
{
"mcpServers": {
"gdp-tools": {
"command": "/absolute/path/to/gdp-mcp-tools/venv/bin/python",
"args": ["/absolute/path/to/gdp-mcp-tools/src/server.py"]
}
}
}
Windows:
{
"mcpServers": {
"gdp-tools": {
"command": "C:\\path\\to\\gdp-mcp-tools\\venv\\Scripts\\python.exe",
"args": ["C:\\path\\to\\gdp-mcp-tools\\src\\server.py"]
}
}
}
To reload: run /mcp reload in Gemini CLI, or restart.
JetBrains (IntelliJ / PyCharm)
Settings > Tools > MCP Servers > Add:
- Name:
gdp-tools - Command:
/absolute/path/to/gdp-mcp-tools/venv/bin/python - Args:
/absolute/path/to/gdp-mcp-tools/src/server.py
Any MCP Client (stdio transport)
/path/to/gdp-mcp-tools/venv/bin/python /path/to/gdp-mcp-tools/src/server.py
Reads JSON-RPC from stdin, writes responses to stdout, logs to stderr.
Credentials
Copy .env.example to src/.env and fill in:
| Credential | Required | How to Get |
|---|---|---|
USERNAME |
Yes | Your LDAP username (auto-detected from $USER if not set) |
JENKINS_TOKEN |
Yes | Jenkins > Profile > Configure > API Token |
OKTA_PASSWORD |
Yes | Your Okta login password |
PRESTO_PASSWORD |
Yes | Trino cluster password (ask team lead) |
JIRA_PERSONAL_TOKEN |
Yes | Jira > Profile > Personal Access Tokens |
JIRA_USERNAME |
Yes | Your @grubhub.com email |
GITHUB_TOKEN |
Optional | GitHub > Settings > Developer Settings > PAT (repo + read:org, SSO for GrubhubProd) |
REDASH_API_KEY |
Optional | dev-redash.gdp.data.grubhub.com > Profile > API Key |
SLACK_MCP_XOXC_TOKEN |
Optional | Slack browser DevTools > Network > Copy from request |
DD_API_KEY |
Optional | DataDog > Org Settings > API Keys |
PAGERDUTY_USER_API_KEY |
Optional | PagerDuty > User Settings > API Token |
DATAHUB_TOKEN |
Optional | DataHub UI > Settings > Access Tokens |
Tool Catalog
Data Platform (13 tools)
| Tool | Description |
|---|---|
sync_table |
Sync prod table to dev |
jenkins_deploy_branch |
Deploy Git branch to dev |
jenkins_deploy |
Deploy version to prod |
jenkins_integrate |
Integrate PR |
jenkins_start_cluster / stop_cluster |
EMR cluster management |
jenkins_validate_schedule |
Validate schedule.json |
azkaban_run_flow |
Execute Azkaban flow |
azkaban_monitor |
Monitor execution until complete |
azkaban_status |
Check execution status |
azkaban_cancel |
Cancel running execution |
azkaban_flows |
List flows in project |
azkaban_list_executions |
List recent executions |
trino_query |
Execute SQL on dev/preprod/prod |
Observability (7 tools)
| Tool | Description |
|---|---|
list_emr_clusters / describe_emr_cluster |
EMR cluster info |
spark_app_details / spark_stage_details |
Spark History Server |
get_flow_logs |
Azkaban execution logs via SSH |
vpn_connect / vpn_status / vpn_disconnect |
F5 VPN management |
s3_list_partitions / s3_list_batch_ids |
S3 data inspection |
Analytics (7 tools)
| Tool | Description |
|---|---|
pv_failure_summary |
PV framework failure analysis |
pv_analyze_metric |
Deep-dive metric investigation |
pv_investigate_root_cause |
Root cause tracing |
pv_generate_report |
HTML report generation |
backfill_plan / backfill_fire / backfill_list_jobs |
Backfill management |
calculate_sync_plan / check_sync_status |
Smart sync planning |
Collaboration (10 tools)
| Tool | Description |
|---|---|
jira_search / jira_get_issue / jira_create_issue |
Jira issues |
jira_transition_issue / jira_add_comment |
Jira workflow |
slack_post / slack_read_message / slack_read_thread / slack_search |
Slack |
github_list_prs / github_read_pr / github_comment_pr / github_update_pr |
GitHub PRs |
get_pr_discussions |
PR discussion threads |
Data Catalog (7 tools)
| Tool | Description |
|---|---|
datahub_search / datahub_get_dataset |
DataHub metadata |
datahub_get_lineage / datahub_get_column_lineage |
DataHub lineage |
datahub_whoami |
DataHub auth check |
redash_run_query / redash_create_query / redash_update_query |
Redash queries |
redash_create_visualization / redash_clone_query |
Redash viz |
create_branded_google_doc / convert_md_to_branded_html |
Google Docs |
File Structure
gdp-mcp-tools/
README.md # This file
TOOLS.md # Detailed tool reference
requirements.txt # Python dependencies
.env.example # Credential template (copy to src/.env)
.mcp.json.example # MCP client config template
.gitignore
src/
server.py # FastMCP entry point (stdio transport)
auth.py # Shared auth (Jenkins, Azkaban, Trino, Jira)
.env # Your credentials (gitignored, never committed)
tools/
jenkins_tools.py # Deploy, sync, integrate, cluster ops
scheduler_tools.py # Azkaban flows, monitor, cancel
trino_tools.py # SQL queries (dev/preprod/prod)
emr_tools.py # EMR cluster list/describe
s3_tools.py # S3 partition listing
backfill_tools.py # Backfill planning + execution
sync_plan_tools.py # Smart sync planning
spark_tools.py # Spark History Server
flowlogs_tools.py # Azkaban log retrieval via SSH
pv_tools.py # PV failure analysis
vpn_tools.py # VPN connect/disconnect/status
jira_tools.py # Jira issue management
slack_tools.py # Slack messaging
github_tools.py # GitHub PR operations
pr_tools.py # PR discussion threads
google_doc_tools.py # Branded Google Doc creation
datahub_tools.py # DataHub lineage/search
redash_tools.py # Redash query/viz management
scripts/ # Utility scripts (not MCP tools)
datadog_wrapper.sh # DataDog MCP wrapper
pagerduty_wrapper.sh # PagerDuty MCP wrapper
google_workspace_wrapper.sh
create_redash_queries.py
config/
config.json # Tool configuration
Troubleshooting
| Problem | Fix |
|---|---|
ModuleNotFoundError: No module named 'mcp' |
pip install mcp[cli] in the venv |
PRESTO_HOST must be set |
Check src/.env exists with correct values |
Connection timed out |
Connect VPN first (vpn_connect tool or manual) |
Azkaban login failed |
Check OKTA_PASSWORD is current |
Jenkins 401 |
Regenerate JENKINS_TOKEN |
Jira 401 |
Regenerate JIRA_PERSONAL_TOKEN |
DNS resolution failed |
VPN dropped, reconnect |
| Tools not showing in client | Restart MCP client. Check .mcp.json paths are absolute. |
| Windows path issues | Use \\ in JSON paths, or forward slashes / |
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。