⏰ Tempo Filler MCP Server

A Model Context Protocol (MCP) server for managing Tempo worklogs in JIRA. This server enables AI assistants to interact with Tempo's time tracking system, allowing for worklog retrieval, creation, bulk operations, and management.

🖼️ Visual UI with MCP Apps

TempoFiller now supports MCP Apps - rendering rich visual interfaces directly in compatible AI chat clients like Claude Desktop. Instead of just text responses, you get interactive calendar and timesheet views!

Timesheet Grid View

Ask for your logged hours and see them in a familiar Tempo-style pivot table:

Features:

• Issues as rows, time periods as columns

• Coverage-aware coloring (green = full, yellow = under, red = gap)

• Zoom toggle: Days / Weeks / Months

• Total row with logged/required hours comparison

Schedule Calendar View

Check your work schedule with a visual calendar:

Features:

• Month grid with working days (green) and non-working days (gray)

• Shows required hours per day

• Holiday and weekend awareness

• Summary with total working days and required hours

Note: Visual UIs render in MCP Apps-compatible hosts (Claude Desktop, VS Code). CLI hosts receive the same structured data, which the AI formats conversationally.

Related MCP server: Jira MCP Server

🚀 Quick Start

Install in VS Code

Install in Claude Desktop

1. Click the button above to download the desktop extension (.dxt file)

2. Open Claude Desktop and go to Settings → Extensions

3. Drag the downloaded .dxt file into the extensions panel to install

   • If dragging doesn't work: Click Advanced Settings → Install Extension and select the downloaded file

4. Fill in the Tempo Base URL and PAT in the environment variables section

5. Don't forget to enable it!

Install Manually on your favorite AI Assistant

{
  "mcpServers": {
    "tempo-filler": {
      "command": "npx",
      "args": ["@tranzact/tempo-filler-mcp-server"],
      "env": {
        "TEMPO_BASE_URL": "https://your-jira-instance.com",
        "TEMPO_PAT": "your-personal-access-token"
      }
    }
  }
}

💬 Real Usage Example

Here's an actual conversation showing how natural and powerful the interaction can be:

👤 User:

fill my october hours with the same project from september

🤖 AI Assistant:

I'll help you fill your October hours with the same project from September. Let me first check your September worklogs to understand the pattern, then fill October accordingly.

🔧 mcp_tempo-filler-_get_worklogs({
  "startDate": "2025-09-01",
  "endDate": "2025-09-30"
})

Perfect! I can see you worked on PROJ-1234 (Tempo MCP Server) for 200 hours in September. Now let me get your October work schedule to see which days need to be filled.

🔧 mcp_tempo-filler-_get_schedule({
  "startDate": "2025-10-01",
  "endDate": "2025-10-31"
})

Excellent! I now have all the information needed. Let me fill all 23 working days in October with the Tempo MCP Server project using 8 hours per working day.

🔧 mcp_tempo-filler-_bulk_post_worklogs({
  "billable": true,
  "worklogs": [
    {"issueKey": "PROJ-1234", "hours": 8, "date": "2025-10-01", "description": "Tempo MCP Server development"},
    {"issueKey": "PROJ-1234", "hours": 8, "date": "2025-10-02", "description": "Tempo MCP Server development"},
    // ... 21 more entries for all working days
  ]
})

✅ October Hours Filled Successfully

• Project: PROJ-1234 (Tempo MCP Server)

• Total Hours: 184 hours across 23 working days

• Hours per Day: 8 hours (matching your standard work schedule)

• Status: All entries created successfully with no failures

Tool Calls Made:

• get_worklogs - Retrieved September pattern (PROJ-1234, 200h)

• get_schedule - Found 23 working days in October (184h capacity)

• bulk_post_worklogs - Created 23 entries in one operation

Your October timesheet is now complete and matches the same project pattern you used in September!

🎯 What Happened Here:

• 3 Tool Calls: get_worklogs → get_schedule → bulk_post_worklogs

• Smart Pattern Recognition: Automatically identified PROJ-1234 from September data

• Schedule Awareness: Only filled working days, skipped weekends automatically

• Bulk Operations: Created 23 worklog entries in a single API call

• Error-Free Execution: 100% success rate with proper validation

• Natural Language: No need to remember issue keys, dates, or tool syntax

⏱️ Time Saved: This single natural language request triggered 3 MCP tool calls and replaced hours of manual timesheet entry!

🛠️ How This Was Built

This MCP server was built in just 3 hours using AI-powered development tools, demonstrating the power of modern AI-assisted coding:

Development Timeline

1. Specification Phase

   • Created the complete technical specification using GitHub Copilot with Claude Sonnet 4

   • Defined all API endpoints, data structures, and tool interfaces

   • Refined requirements through iterative conversation

2. Implementation Phase

   • Used VS Code with Claude Code to one-shot the entire implementation

   • Generated complete TypeScript codebase, tool implementations, and client logic

   • Implemented all core functionality in a single AI-assisted session

3. Refinement Phase

   • Switched back to GitHub Copilot with Claude Sonnet 4 after hitting usage limits in Claude Code

   • Fixed API payload formatting and authentication issues

   • Debugged and polished the Tempo API integration

Key Success Factors

• Clear specification first: Having a detailed spec enabled effective one-shot implementation

• AI tool synergy: Different AI tools excelled at different phases of development

• Iterative refinement: Quick feedback loops with AI assistants for debugging

This project showcases how AI-powered development can dramatically accelerate the creation of robust, production-ready tools.

✨ Features

• Get Worklogs: Retrieve worklogs for users with date range and issue filtering

• Create Worklogs: Add single worklog entries with automatic issue resolution

• Bulk Operations: Create multiple worklog entries efficiently using concurrent processing

• Delete Worklogs: Remove existing worklog entries

• Get Schedule: Retrieve work schedule with working/non-working day information

• Visual UIs: Rich calendar and timesheet grid views via MCP Apps

• Resource Access: Browse worklog data and recent issues

• Prompt Templates: Generate analysis prompts for worklog data

📦 Installation

Prerequisites

• Node.js (version 18 or higher)

• A JIRA instance with Tempo Timesheets plugin installed

• Personal Access Token for your JIRA account

NPX (Recommended)

The easiest way to use the server is with npx - no installation required:

npx @tranzact/tempo-filler-mcp-server

Just configure your AI assistant to use npx @tranzact/tempo-filler-mcp-server as the command.

Development Setup (Source)

For development or customization:

1. Clone the repository:

   git clone https://github.com/TRANZACT/tempo-filler-mcp-server
   cd TempoFiller

2. Install dependencies and build:

   npm install && npm run build

⚙️ Configuration

The server requires environment variables for authentication and configuration:

Required Environment Variables

• TEMPO_BASE_URL: Your JIRA instance URL (e.g., https://jira.company.com)

• TEMPO_PAT: Personal Access Token for authentication

Optional Environment Variables

• TEMPO_DEFAULT_HOURS: Default hours per workday (default: 8)

Creating a Personal Access Token (PAT)

1. Log into your JIRA instance

2. Go to Profile → Personal Access Tokens

3. Click Create token

4. Give it a name (e.g., "Tempo MCP Server")

5. Set appropriate permissions (read/write access to issues and worklogs)

6. Copy the token value for use in TEMPO_PAT

🛠️ Available Tools

1. get_worklogs - Retrieve Time Logs

Retrieve worklogs for a date range with optional filtering. In MCP Apps-compatible hosts, displays an interactive timesheet grid.

Parameters:

• startDate (string): Start date in YYYY-MM-DD format

• endDate (string, optional): End date, defaults to startDate

• issueKey (string, optional): Filter by specific issue key

Visual Output (MCP Apps):

Example Usage:

"Get my July hours"
→ Returns: Total: 184h (23 entries)
          • PROJ-1234: 184.0h (23 entries)

"Show me my worklogs for PROJ-1234 in July"
→ Filters results to specific issue

2. post_worklog - Log Single Entry

Create a new worklog entry for a specific issue and date.

Parameters:

• issueKey (string): JIRA issue key (e.g., "PROJ-1234")

• hours (number): Hours worked (decimal, 0.1-24)

• startDate (string): Date in YYYY-MM-DD format

• endDate (string, optional): End date for multi-day entries

• billable (boolean, optional): Whether time is billable (default: true)

• description (string, optional): Work description

Example Usage:

"Log 8 hours to PROJ-1234 for July 10th"
→ Returns: ✅ Worklog Created Successfully
          Issue: PROJ-1234 - Example Project Task
          Hours: 8h
          Date: 2025-07-10
          Worklog ID: 1211549

3. bulk_post_worklogs - Create Multiple Entries

Create multiple worklog entries efficiently with concurrent processing.

Parameters:

• worklogs (array): Array of worklog objects:

  • issueKey (string): JIRA issue key

  • hours (number): Hours worked

  • date (string): Date in YYYY-MM-DD format

  • description (string, optional): Work description

• billable (boolean, optional): Whether time is billable for all entries

Example Usage:

"Post 8 hours a day every weekday from July 11 to 15 on PROJ-1234"
→ Returns: ✅ Bulk Worklog Creation Started
          Processing 3 worklog entries...
          ✅ Successful: 3
          ❌ Failed: 0
          📊 Total Hours: 24

"Fill all weekdays in July with 8 hours on PROJ-1234"
→ Creates 23 entries for all weekdays in the month

4. delete_worklog - Remove Entry

Delete an existing worklog entry by ID.

Parameters:

• worklogId (string): Tempo worklog ID to delete

Example Usage:

"Delete worklog with ID 1211547"
→ Removes the specified worklog entry

5. get_schedule - Retrieve Work Schedule

Retrieve work schedule information showing working days, non-working days, and expected hours per day. In MCP Apps-compatible hosts, displays an interactive calendar view.

Parameters:

• startDate (string): Start date in YYYY-MM-DD format

• endDate (string, optional): End date, defaults to startDate

Visual Output (MCP Apps):

Example Usage:

"What's my work schedule for October 2025?"
→ Returns: ## Work Schedule (2025-10-01 to 2025-10-31)

          **Period Summary:**
          - Total Days: 31
          - Working Days: 23
          - Non-Working Days: 8
          - Total Required Hours: 184h

"Show me working days this week"
→ Returns schedule for current week with working/non-working day breakdown

💡 Integration Tip: Use this tool before bulk worklog operations to identify working days and avoid logging time on weekends/holidays.

🌎 Real-World Usage Examples

Based on successful implementation, here are practical scenarios:

Daily Time Logging

# Using VS Code with GitHub Copilot:
User: "Log 8 hours on PROJ-1234 for today"
AI: ✅ Worklog Created Successfully
    Issue: PROJ-1234 - Example Project Task
    Hours: 8h
    Date: 2025-08-01
    Worklog ID: 1211550
    Time Spent: 1d

Schedule-First Bulk Time Filling

# RECOMMENDED: Check schedule first, then create entries:
User: "Check my October schedule, then fill all working days with 8 hours on PROJ-1234"
AI: ## Work Schedule (2025-10-01 to 2025-10-31)
    **Period Summary:**
    - Working Days: 23
    - Non-Working Days: 8
    - Total Required Hours: 184h

    ✅ Bulk Worklog Creation Started
    Processing 23 worklog entries (working days only)...
    ✅ Successful: 23
    ❌ Failed: 0
    📊 Total Hours: 184

Monthly Time Filling (Legacy)

# Direct bulk operation (may include non-working days):
User: "Fill all weekdays in July with 8 hours on PROJ-1234"
AI: ✅ Bulk Worklog Creation Started
    Processing 23 worklog entries...
    ✅ Successful: 23
    ❌ Failed: 0
    📊 Total Hours: 184

Schedule-Aware Planning

# Smart time planning workflow:
User: "What's my capacity this month and how should I distribute my time?"
AI: ## Work Schedule (2025-10-01 to 2025-10-31)
    - Working Days: 23
    - Total Required Hours: 184h
    - Average Daily Hours: 8h

    **💡 Next Steps - Schedule-Aware Time Logging:**
    - Total capacity is 184h across 23 working days
    - Example: "Fill all working days shown above with 8 hours on PROJ-1234"

Time Tracking Analysis

# Monthly summary:
User: "Get my July hours"
AI: 📊 Total Hours: 184 hours (23 entries)

    Breakdown by issue:
    • PROJ-1234: 184.0h (23 entries)

    Daily pattern: 8 hours per weekday
    Completion: 100% (all weekdays filled)

🤖 Development

Project Structure

src/
├── index.ts              # Main MCP server entry point (stdio transport)
├── http-server.ts        # HTTP transport server (development/testing)
├── tempo-client.ts       # Tempo API client with PAT auth
├── tools/                # Tool implementations
│   ├── get-worklogs.ts
│   ├── post-worklog.ts
│   ├── bulk-post.ts
│   ├── delete-worklog.ts
│   └── get-schedule.ts
├── types/                # TypeScript type definitions
│   ├── tempo.ts          # Tempo API types
│   ├── mcp.ts            # MCP validation schemas
│   ├── responses.ts      # JSON response types
│   └── index.ts
└── ui/                   # MCP Apps visual components
    ├── get-schedule/     # Calendar view
    │   ├── index.html
    │   ├── index.ts
    │   └── styles.css
    └── get-worklogs/     # Timesheet grid view
        ├── index.html
        ├── index.ts
        └── styles.css

Build Commands

• npm run build: Compile TypeScript + build UI bundles + create MCP bundle

• npm run build:ui: Build UI components only (Vite)

• npm run dev: Build and run the server (stdio)

• npm run dev:http: Build and run HTTP server (for MCP Apps testing)

• npm run typecheck: Type checking without compilation

License

ISC License - see package.json for details

Contributing

Contributions are welcome! Please follow the existing code style and ensure all tools work correctly with real Tempo API endpoints.