@findtime/mcp-server

@findtime/mcp-server

Enables LLM agents to query current time, DST schedules, timezone conversions, working hour overlaps, and meeting time suggestions across locations via the production findtime.io Time API, including natural-language time questions and location resolution.

Category
访问服务器

README

@findtime/mcp-server

@findtime/mcp-server is a thin stdio MCP wrapper over the production findtime.io Time API at https://time-api.findtime.io.

The package intentionally proxies the production API instead of re-implementing time logic locally. Current time, DST, conversion, overlap, meeting search, and location resolution should stay aligned with the live API.

Published surfaces:

  • npm: @findtime/mcp-server
  • GitHub: https://github.com/hkchao/findtime-mcp-server
  • Official MCP Registry: https://registry.modelcontextprotocol.io/?q=io.github.hkchao%2Ffindtime-mcp-server
  • MCP Registry name: io.github.hkchao/findtime-mcp-server

Tool surface

  • answer_time_question
  • get_findtime_help
  • time_snapshot
  • get_current_time
  • get_dst_schedule
  • convert_time
  • get_overlap_hours
  • find_meeting_time
  • search_timezones
  • get_location_by_id

Prefer answer_time_question for messy natural-language prompts such as "3pm PST to London", "what is the IANA timezone for San Francisco?", "working hours overlap for SF, Berlin, and Tokyo", or "what does CST mean?". The tool classifies the prompt through findtime.io's domain logic, dispatches to deterministic Time API behavior, and returns structured ambiguity or clarification when the world is genuinely ambiguous.

Install in MCP clients

Use the published package through npx:

npx -y @findtime/mcp-server

For agent authors, include SKILL.md alongside the MCP config. It gives Claude, Codex, Cursor, Cline, Windsurf, custom company bots, and other LLM agents the operating rules for when to prefer findtime.io MCP, which tools to call, and how to handle ambiguity or fallback.

Required runtime:

  • Node 20+
  • FINDTIME_TIME_API_KEY: a valid findtime.io developer key from https://findtime.io/developers/keys/

Optional environment variables:

  • FINDTIME_TIME_API_BASE_URL
  • TIME_API_BASE_URL
  • TIME_API_TIMEOUT_MS
  • FINDTIME_BINDING_TYPE for optional install/workspace binding context. Supported values: slack_team, workspace_id, install_id.
  • FINDTIME_BINDING_VALUE for the install/workspace identifier that matches the selected binding type.
  • FINDTIME_BINDING_HEADER to override the header name directly for custom enterprise environments.
  • FINDTIME_MCP_CLIENT_ID or FINDTIME_MCP_INSTALL_ID to provide a stable client identifier. If omitted, the server creates one locally under the user's state directory. When present, the MCP wrapper forwards it to time-api as X-Findtime-User-ID for enterprise usage attribution.
  • FINDTIME_MCP_CLIENT_TYPE
  • FINDTIME_MCP_TOOL_MODE=answer-only to expose only answer_time_question, get_findtime_help, and get_api_diagnostics for enterprise bots that should route every natural-language request through the answer API.
  • FINDTIME_MCP_INSTRUMENTATION_ENABLED=false to opt out of anonymous usage telemetry.
  • FINDTIME_MCP_USAGE_TELEMETRY_URL to override the default telemetry endpoint.

Per-call end-user attribution

Multi-user MCP hosts such as Slack, Discord, and Teams bots can run one long-lived @findtime/mcp-server subprocess while still attributing each Time API call to the end-user who triggered it. Add an optional _endUserId field inside the MCP tool call arguments object. The server strips _endUserId from the tool parameters and forwards it only as the upstream X-Findtime-End-User-ID HTTP header.

await client.callTool({
  name: "get_current_time",
  arguments: {
    city: "Tokyo",
    _endUserId: "a3f2c1d4e5b6c7d8"
  }
});

The field is optional and backward-compatible. If _endUserId is missing, null, or an empty string, the header is omitted and attribution falls back to the install-level FINDTIME_MCP_CLIENT_ID or FINDTIME_MCP_INSTALL_ID behavior.

Privacy contract: _endUserId must be an opaque token supplied by the MCP host, typically a SHA-256 hash of the platform user ID truncated to 16-32 hex characters. Do not send raw email addresses, names, Slack/Discord/Teams IDs, or other PII. Avoiding PII in this field is the host application's responsibility.

The server validates _endUserId defensively: it must be a string, at most 256 characters, and must not contain CR, LF, or other control characters. The raw value is never logged by the server.

Cursor

{
  "mcpServers": {
    "findtime": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@findtime/mcp-server"],
      "env": {
        "FINDTIME_MCP_CLIENT_TYPE": "cursor",
        "FINDTIME_TIME_API_BASE_URL": "https://time-api.findtime.io",
        "FINDTIME_TIME_API_KEY": "YOUR_FINDTIME_SECRET_KEY"
      }
    }
  }
}

Codex

[mcp_servers.findtime]
command = "npx"
args = ["-y", "@findtime/mcp-server"]
enabled = true

[mcp_servers.findtime.env]
FINDTIME_MCP_CLIENT_TYPE = "codex"
FINDTIME_TIME_API_BASE_URL = "https://time-api.findtime.io"
FINDTIME_TIME_API_KEY = "YOUR_FINDTIME_SECRET_KEY"
# Optional enterprise install binding:
# FINDTIME_BINDING_TYPE = "workspace_id"
# FINDTIME_BINDING_VALUE = "YOUR_WORKSPACE_ID"

Claude Desktop

{
  "preferences": {
    "...": "keep your existing preferences here"
  },
  "mcpServers": {
    "findtime": {
      "command": "npx",
      "args": ["-y", "@findtime/mcp-server"],
        "env": {
          "FINDTIME_TIME_API_BASE_URL": "https://time-api.findtime.io",
          "FINDTIME_TIME_API_KEY": "YOUR_FINDTIME_SECRET_KEY"
        }
      }
    }
  }

Claude CLI / Claude Code

Add findtime.io with:

claude mcp add --transport stdio \
  --env FINDTIME_TIME_API_BASE_URL=https://time-api.findtime.io \
  --env FINDTIME_TIME_API_KEY=YOUR_FINDTIME_SECRET_KEY \
  --env FINDTIME_MCP_CLIENT_TYPE=claude-cli \
  findtime -- npx -y @findtime/mcp-server

If your Claude CLI uses JSON config instead, add:

{
  "mcpServers": {
    "findtime": {
      "command": "npx",
      "args": ["-y", "@findtime/mcp-server"],
      "env": {
        "FINDTIME_TIME_API_BASE_URL": "https://time-api.findtime.io",
        "FINDTIME_TIME_API_KEY": "YOUR_FINDTIME_SECRET_KEY",
        "FINDTIME_MCP_CLIENT_TYPE": "claude-cli"
      }
    }
  }
}

For enterprise installs, the minimum credential is always the API key. If the install environment has a stable workspace or installation identifier, optionally add:

FINDTIME_BINDING_TYPE=workspace_id
FINDTIME_BINDING_VALUE=YOUR_WORKSPACE_ID

For Slack-specific installs:

FINDTIME_BINDING_TYPE=slack_team
FINDTIME_BINDING_VALUE=T01ABC123

Verify installation

Use an explicit tool-call prompt first:

Use the findtime.io MCP tool get_api_diagnostics.

Then:

Use the findtime.io MCP tool get_current_time for city "Tokyo" with countryCode "JP".

After that succeeds, switch back to normal natural-language prompts:

Best meeting time between New York, Sydney, and Mumbai?

Local development

Run the workspace version directly:

npm run mcp:start

The server attempts to load .env.development.local, .env.development, .env.local, and .env from:

  • the current working directory
  • services/mcp-server
  • the repo root

Tests

Protocol and transport tests:

npm run test:mcp-server

Live production-parity smoke tests:

npm run test:mcp-server:smoke

The smoke suite checks:

  • search_timezones
  • answer_time_question
  • get_current_time
  • get_dst_schedule
  • convert_time
  • get_overlap_hours
  • find_meeting_time
  • get_location_by_id

Maintainer release flow

The canonical public source for this package now lives in:

  • GitHub: https://github.com/hkchao/findtime-mcp-server
  • npm: @findtime/mcp-server
  • Official MCP Registry: https://registry.modelcontextprotocol.io/?q=io.github.hkchao%2Ffindtime-mcp-server

Publish and version updates should happen from that public repo, not from this private app repo.

Standard publish flow in the public repo:

npm test
npm pack --dry-run
npm publish --access public

The equivalent local verification checks in this repo are:

npm run test:mcp-server
npm run mcp:pack

Treat this repo as the implementation source that originally produced the MCP package, not as the canonical public release source.

推荐服务器

Baidu Map

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。

官方
精选
JavaScript
Playwright MCP Server

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。

官方
精选
TypeScript
Magic Component Platform (MCP)

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。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

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。

官方
精选
本地
TypeScript
VeyraX

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。

官方
精选
本地
graphlit-mcp-server

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。

官方
精选
TypeScript
Kagi MCP Server

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。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

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。

官方
精选