timepiece-mcp
MCP server integrating OBSS Timepiece Time in Status for Jira, enabling queries about issue durations, status histories, and aggregated metrics via natural language.
README
timepiece-mcp
An MCP server that lets Claude query OBSS Timepiece Time in Status data for Jira issues.
Ask Claude things like:
- "How long has PROJ-123 spent in each status?"
- "Show me the full lifecycle of PROJ-456 including how many times it went back into Development"
- "What is the average cycle time for tickets closed this sprint?"
- "Which status is causing the biggest delays in our project?"
- "Export a CSV of time-in-status for all tickets resolved this month"
Prerequisites
- uv — fast Python package manager
- A Jira instance with Timepiece installed
- A personal Timepiece API token (see below)
Get your Timepiece API token
- In Jira, go to Apps → Timepiece → API Settings
- Click Create New Token
- Copy the token
⚠️ Keep this token secret — it grants access to your Jira data through Timepiece.
Installation
Claude Desktop
Add this to ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"timepiece": {
"command": "uvx",
"args": ["timepiece-mcp"],
"env": {
"TIMEPIECE_TOKEN": "your-token-here",
"TIMEPIECE_CALENDAR": "your-calendar-id"
}
}
}
}
Restart Claude Desktop. You'll see a 🔌 icon confirming the server is connected.
Claude Code (CLI)
claude mcp add timepiece uvx timepiece-mcp \
-e TIMEPIECE_TOKEN=your-token-here \
-e TIMEPIECE_CALENDAR=your-calendar-id
Environment variables
| Variable | Required | Default | Description |
|---|---|---|---|
TIMEPIECE_TOKEN |
✅ Yes | — | Your personal Timepiece API token |
TIMEPIECE_CALENDAR |
Recommended | — | Default calendar ID for business-hours calculations. Without this, durations use a 24h/day UTC clock and will be wrong for most teams. |
TIMEPIECE_DEFAULT_DAY_LENGTH |
No | businessDays |
businessDays or calendarDays |
TIMEPIECE_DEFAULT_VIEW_FORMAT |
No | days |
days, hours, minutes, or seconds |
Finding your calendar ID: In Jira go to Apps → Timepiece → Settings → Calendars. The numeric ID is shown next to each calendar. You can also ask Claude to run timepiece_list_calendars after setup to see all available calendars and their IDs.
Available tools
| Tool | What it does |
|---|---|
timepiece_get_issue |
Time spent in each status for a single issue — returned as a Status / Duration / Visits table |
timepiece_get_issue_expanded |
Full status history for a single issue: visit counts, min/max/average per status, and a chronological transition timeline |
timepiece_list_issues |
Time-in-status for a set of issues defined by a JQL query, returned as an issues × statuses matrix |
timepiece_aggregate |
Average, median, sum or standard deviation of time-in-status across any JQL-filtered set of issues; also supports lead time and cycle time calculations |
timepiece_list_calendars |
Lists all Timepiece calendars configured for your Jira instance, including IDs, timezones and working hours |
timepiece_search_calendar |
Finds a calendar by name and returns its ID |
timepiece_export_sync |
Exports a time-in-status report as CSV or XLSX for any JQL-filtered set of issues, saving the file locally |
Example prompts
How long did PROJ-123 spend in each status?
Show me the full lifecycle of PROJ-456 including how many times it went back into Development
What is the average cycle time for tickets closed in PROJ this sprint?
Which status is causing the biggest delays in project PROJ?
Compare average development time across the team for last sprint
Export a CSV of time-in-status for all PROJ tickets resolved this month
What calendars are configured in Timepiece?
Sharing with your team
Each team member needs their own Timepiece token:
- Get a token from Jira → Apps → Timepiece → API Settings
- Add the config block above to their Claude Desktop or Claude Code config
- Use their own token — tokens are personal and tied to each person's Jira identity
Local development
git clone https://github.com/gianni1989/timepiece-mcp
cd timepiece-mcp
cp .env.example .env # fill in your token and calendar ID
uv sync
uv run timepiece-mcp
Run the test suite:
TIMEPIECE_TOKEN=your-token TIMEPIECE_CALENDAR=your-calendar-id uv run python test_tools.py
Troubleshooting
"TIMEPIECE_TOKEN environment variable is not set"
→ Add TIMEPIECE_TOKEN to the env block in your Claude config and restart.
"Unauthorised — your TIMEPIECE_TOKEN is invalid or expired" → Go to Jira → Apps → Timepiece → API Settings and generate a new token.
"Not found — the issue key does not exist"
→ Verify the issue key format: it must be PROJECT-NUMBER in uppercase, e.g. PROJ-123.
Durations seem too large (e.g. 3× bigger than expected)
→ Set TIMEPIECE_CALENDAR to your calendar's numeric ID. Without it, the API uses a 24h/day UTC clock instead of your business hours.
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。