Temporal MCP Server
An MCP server for interacting with Temporal Cloud workflows across multiple regions, providing tools to list, describe, terminate, and retrieve step results from workflows via natural language.
README
Temporal MCP Server
A Model Context Protocol (MCP) server for Temporal Cloud that enables Claude Code to interact with your Temporal workflows across all regions.
š Multi-Region Support
Access different Temporal Cloud regions:
- šŗšø US (us-west-2)
- šŖšŗ EU (eu-central-1)
- š¦šŗ AU (ap-southeast-2)
- š®š³ IN (ap-south-1)
š Quick Setup
1. Configure Environment
Copy .env.example and setup the values
2. Add to Claude Code
Edit ~/.claude.json:
{
"mcpServers": {
"temporal": {
"command": "node",
"args": ["/absolute/path/to/your/temporal-mcp/build/index.js"],
"env": {
"TEMPORAL_API_KEY": "your-actual-api-key",
"TEMPORAL_ACCOUNT_ID": "<cloud_account_id>",
"DEFAULT_REGION": "us"
}
}
}
}
3. Restart Claude Code
That's it! The server auto-connects to any region based on your queries.
š ļø Available Tools
1. get_workflow_step_results ā (Recommended)
Get processed step results with resultUri for downloading logs - use this instead of full history!
Use when: You want step summaries, resultUri, durations, or to debug flows Returns: Compact JSON with step names, resultUri, success status, durations
"Get step results for workflow 'refresh-monitors-456' run 'abc-123' in EU region"
Response:
{
"success": true,
"totalSteps": 3,
"steps": [
{
"stepName": "fetchData",
"isSuccess": true,
"resultUri": "flowData/99/99999/flow-id/run-id/step-fetchData/1234567890.json",
"durationSeconds": "2.45",
"exposedData": { "count": 150 }
}
]
}
2. list_workflow_executions
Query workflows with filters using Temporal visibility syntax.
{
"query": "WorkflowType='refreshSlackUsersFlow' AND ExecutionStatus='Running'",
"pageSize": 20,
"region": "eu"
}
Example queries:
"ExecutionStatus='Running'"- All running workflows"ExecutionStatus='Failed' AND StartTime > '2026-02-01T00:00:00Z'"- Recent failures"WorkflowType='refreshGithubUsersFlow'"- Specific workflow type
3. describe_workflow_execution
Get workflow status, timing, and metadata.
{
"workflowId": "my-workflow-123",
"region": "us"
}
4. terminate_workflow_execution
Stop a running workflow.
{
"workflowId": "stuck-workflow",
"reason": "Manual termination - investigating data source issue",
"region": "au"
}
5. list_closed_workflow_executions
List workflows that completed in a time range.
{
"startTime": "2026-02-01T00:00:00Z",
"endTime": "2026-02-05T23:59:59Z",
"region": "in"
}
6. get_workflow_execution_history
Get raw Temporal event history - only use if you need full history!
Warning: Can be 80k+ characters. Use get_workflow_step_results instead.
š” Usage Examples
Get Step Results with ResultUri (Most Common)
"Use temporal MCP and get step results for workflow 'intune.refreshIntuneEntitiesFlow.orgPk_756d...' run 'abc-123' and show me the resultUri for each step"
Debug Failed Flow
"Show me failed workflows in EU region from last 24 hours, then get step results for the most recent failure"
Monitor Specific Flow
"Is the GitHub user sync running for org 12345 in US region?"
Cross-Region Health Check
"Compare total running workflows across all regions"
Performance Analysis
"Get step results for the last 5 runs of 'refreshSlackUsersFlow' in US region and show me average step durations"
š Region Handling
All tools accept an optional region parameter:
{ "region": "us" } // United States (default)
{ "region": "eu" } // Europe
{ "region": "au" } // Australia
{ "region": "in" } // India
Without region: Uses DEFAULT_REGION from config
Natural language: Claude understands "EU region", "Australia", etc.
Automatic Mapping
The server automatically maps regions to endpoints No manual configuration needed!
š§ Technical Details
What get_workflow_step_results Does
- Fetches workflow history from Temporal
- Filters for activity completed/failed events
- Decodes base64 payloads ā JSON
- Extracts stepName, resultUri, exposedData
- Calculates step durations
- Deduplicates by step name
- Returns compact summary
Connection Management
- Lazy-loads connections (only connects when needed)
- Maintains separate connection per region
- Reuses connections automatically
- First query to a region: ~1-2 seconds
- Subsequent queries: Fast (cached connection)
Event Types Processed
- Event Type 12: Activity Task Completed
- Event Type 13: Activity Task Failed
- Event Type 3: Workflow Execution Failed
šÆ Best Practices
ā Do This
- Use
get_workflow_step_resultsfor most queries - Specify region when known
- Use natural language with Claude
- Download logs via resultUri
ā Avoid This
- Using
get_workflow_execution_historyunless needed - Parsing raw history manually
- Forgetting to specify region for known workflows
š Common Patterns
Pattern 1: Debugging Failed Flows
1. "List failed flows from last hour in US region"
2. "Get step results for workflow XYZ run ABC"
3. Use resultUri to download logs
4. Analyze error from step data
Pattern 2: Performance Investigation
1. "Get step results for last 5 successful runs"
2. "Compare step durations"
3. "Identify slowest steps"
Pattern 3: Cross-Region Monitoring
1. "Query all regions for specific workflow type"
2. "Aggregate results by region"
3. "Identify regional differences"
š Troubleshooting
"Can't find workflow"
ā Try searching other regions or verify workflow ID
"Connection error"
ā Check API key and account ID in config
"Response too large"
ā Use get_workflow_step_results instead of full history
"Region not working"
ā Verify region code is one of: us, eu, au, in
š Response Format
All responses include region information:
{
"success": true,
"region": "eu",
"namespace": "your_temporal_host.<cloud_account_id>",
"data": { ... }
}
Step results also include:
{
"totalSteps": 3,
"steps": [
{
"stepName": "...",
"isSuccess": true,
"resultUri": "...",
"durationSeconds": "2.45",
"exposedData": { ... }
}
]
}
š¦ Getting Started
- Add your API key to
.env - Update Claude Code config with paths and credentials
- Restart Claude Code
- Test:
"List workflows in US region" - Get step results:
"Get step results for workflow X run Y" - Use resultUri to download logs
š Quick Examples
Simple:
"List running workflows"
With region:
"Show failed workflows in EU from today"
Get step results:
"Get step results for workflow X run Y and show me resultUri"
Cross-region:
"Search all regions for workflow containing 'github'"
Performance:
"Compare execution times between US and EU regions"
Ready to use! Just add your API key and start debugging workflows with Claude! š
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.