LTD MCP

io.github.Eshller由 GitHub 账号 Eshller 发布v0.1.0更新于 Oct 24, 2026

Project management MCP for AI agents with safe task reads and writes.

已验证Streamable HTTP可网页运行效率与工作流数据与分析

概览

AI 生成的概览

远程 MCP 服务器，让 AI 助手读取和写入 LTD 工作区中的项目工作：项目、任务、评论和分析数据。

可以做什么
列出并描述可访问的项目，包括状态、自定义字段和成员；列出令牌持有者的未完成任务，读取任务评论和完整任务详情；搜索和筛选任务；获取项目健康汇总，运行分析聚合并下钻到数据点；通过两步预览-确认流程创建任务、更新任务字段和添加评论。
适用场景
当团队使用 LTD 工作区，并希望助手检查项目健康状况、查找或汇总任务，或在人工确认下起草并提交任务变更和评论时使用。
运行要求
远程 Streamable HTTP 端点，无需本地安装包。需要在 LTD 设置中创建 LTD API 令牌，并以 Bearer 令牌形式放在 Authorization 请求头中。只读访问需要 mcp:read 范围，创建任务、更新、指派和评论需要 mcp:write 范围。需要能访问该端点的网络。
安装前请注意
Authorization 令牌可访问你的 LTD 工作区，请妥善保密，并在 LTD 设置中撤销或轮换。写入工具会创建、更新真实项目数据并添加评论，确认前请逐条检查预览；确认令牌短期有效且只能使用一次。该服务器处于早期公开预览阶段，行为可能变化。

安装

在 SourceWeft 中

  1. 打开 控制台中的 LTD MCP，将其添加到工作区。
  2. 连接 Authorization token 凭据。它们会被私密存储，绝不会在公开页面上展示。
  3. 为需要使用其工具的对话启用该服务。

Web executable，通过 Streamable HTTP。 远程服务在工作区中配置后即可从网页运行时运行。

交给智能体安装

把这段提示词粘贴到 Claude Code、Codex、Cursor 等智能体，或 SourceWeft 对话中。智能体会读取该服务器的安装指南，先向你展示发布者、端点和登录方式，经你同意后再把它添加到你的客户端。密钥不会写进任何命令。

阅读 https://sourceweft.com/mcp/io.github.Eshller%2Fltd-mcp/install.md，并按其中说明把这个 MCP 服务器接入我的客户端。修改任何配置之前，先告诉我它的发布者、端点和登录方式，等我确认。不要把 API 密钥写进命令或我会提交的文件；问我密钥应从哪里获取。

查看智能体所依据的安装指南

其他 MCP 客户端

把它添加到你客户端的 mcpServers 配置中。

{
  "mcpServers": {
    "ltd": {
      "type": "http",
      "url": "https://api.eshway.com/mcp/",
      "headers": {
        "Authorization": "Bearer <YOUR_TOKEN>"
      }
    }
  }
}

README

LTD MCP

Project management MCP for AI agents.

LTD MCP lets AI agents connect to an LTD workspace over MCP and safely operate project work: inspect project health, search tasks, create tasks, update tasks, assign owners, and add comments using scoped API tokens and preview-confirm writes.

Status: early public preview. Read tools are available. Write tools use a two-step preview-confirm safety flow.

Why this exists

Most project management tools were built for humans clicking dashboards. AI agents need safe, structured tools.

LTD MCP is the agent-native interface for LTD, an anti-Jira-style project workspace for teams that want humans and AI agents operating together.

MCP endpoint

txt
https://api.eshway.com/mcp/

Transport: Streamable HTTP / HTTP MCP

Authentication: Bearer token

txt
Authorization: Bearer <LTD_API_TOKEN>

Capabilities

Read tools

  • ltd_list_projects - list projects the token can access
  • ltd_describe_project - describe statuses, custom fields, and members
  • ltd_search_tasks - search/filter tasks
  • ltd_get_task - fetch full task detail
  • ltd_my_tasks - list token owner's open tasks
  • ltd_get_comments - read task comments
  • ltd_project_pulse - project health rollups
  • ltd_analytics_query - analytics aggregations
  • ltd_drilldown_tasks - drill into analytics datapoints

Write tools

  • ltd_create_task - create a task
  • ltd_update_task - update fields on a task
  • ltd_add_comment - add a comment

Write tools are intentionally two-step:

  1. mode="preview" returns the exact proposed change and a short-lived confirm token. Nothing is written.
  2. mode="confirm" with the same arguments plus confirm_token performs the write.

Token scopes

Create an LTD API token from LTD Settings → API tokens.

Recommended scopes:

  • mcp:read for read-only project visibility
  • mcp:write for task creation, updates, assignment, and comments

Keep tokens secret. Revoke and rotate them from LTD settings when needed.

Quickstart

Generic MCP HTTP config

json
{  "mcpServers": {    "ltd": {      "transport": "http",      "url": "https://api.eshway.com/mcp/",      "headers": {        "Authorization": "Bearer ${LTD_API_TOKEN}"      }    }  }}

Hermes

bash
mcp add ltd --transport http \  --url https://api.eshway.com/mcp/ \  --header "Authorization: Bearer $LTD_API_TOKEN"

Smithery

bash
smithery mcp add https://api.eshway.com/mcp/ --id ltd

Example prompts

txt
Show me all LTD projects I can access.
txt
Give me a project pulse for PS - Mini and list the top risks.
txt
Create a low-priority test task in PS - Mini assigned to Eshu, but show me the preview before confirming.
txt
Find stale tasks assigned to me, group them by status, and suggest next actions.

Safety model

LTD MCP is designed for human-controlled agent operations:

  • Scoped PATs limit access.
  • Read and write scopes are separate.
  • Write actions use preview-confirm.
  • Preview explicitly returns nothing_changed_yet: true.
  • Confirm tokens are short-lived and single-use.
  • Invalid assignees/statuses are rejected before write.

Example write flow

Preview:

json
{  "project_id": 168,  "title": "Test task created by ZERO",  "description": "Verify LTD MCP task creation.",  "status": "To Do",  "assignees": ["Eshu"],  "priority": "Low",  "mode": "preview"}

Confirm:

json
{  "project_id": 168,  "title": "Test task created by ZERO",  "description": "Verify LTD MCP task creation.",  "status": "To Do",  "assignees": ["Eshu"],  "priority": "Low",  "mode": "confirm",  "confirm_token": "<token from preview>"}

Supported clients

Any MCP client that supports remote HTTP / Streamable HTTP MCP servers should work.

Planned docs:

  • Claude Desktop
  • Cursor
  • Cline
  • Windsurf
  • Hermes / ZERO
  • Smithery

Links

Roadmap

  • Public hosted docs
  • Demo workspace
  • Client-specific setup guides
  • Official MCP Registry listing
  • Smithery listing
  • MCP.so listing
  • Audit log docs
  • OAuth support

License

MIT for documentation and examples in this repository. LTD server implementation and hosted service are owned by Eshway.

来源：README.md，提交 bb04aad

工具

0
该服务器需要 API 密钥等凭据，因此 SourceWeft 无法在没有凭据的情况下读取其工具，也未收录。

版本历史

1
  1. v0.1.0最新Sep 16, 2026

评分

已满足适用检查项的 85%

该评分衡量此收录条目的完整程度，依据注册中心条目、端点与公开来源计算得出。它不是对服务器本身的评价，也不表示服务器安全或当前可用：可连通性是带日期的检查结果，评分也不包含安装量或用户评价。

  • 近 30 天内可连通必需25 分
  • 已声明认证方式必需10 分
  • 已列出工具15 分

    请让服务器在无凭据时响应 tools/list，或发布可在沙箱中无凭据启动的软件包。

  • 描述必需15 分
  • 图标10 分
  • 文档15 分
  • 许可证10 分