Astrotask
效率与工作流by marktoda
Manage and organize tasks efficiently with AI agent integration. Create, update, query, and track tasks with hierarchical support and real-time feedback. Enhance productivity by leveraging structured task management tools designed for seamless AI interaction.
什么是 Astrotask?
Manage and organize tasks efficiently with AI agent integration. Create, update, query, and track tasks with hierarchical support and real-time feedback. Enhance productivity by leveraging structured task management tools designed for seamless AI interaction.
README
Astrotask

Task manager built for humans and AI agents
Offline‑ready · MCP‑compatible · Fully‑type‑safe · Extensible
Why Astrotask?
Astrotask keeps workstreams organised and agent‑ready:
- Stay on track – persistent workspace records every decision and status change.
- Parallel agent orchestration – multiple agents can tackle non‑blocking tasks simultaneously; the dependency graph prevents collisions.
- Rich, evolving context – attach arbitrary metadata, comments, and notes to improve recall and quality of work.
- Situational awareness – agents can refetch the original task prompt at any time and understand where it fits in the wider roadmap.
- Offline‑first – capture ideas without a network and sync when back online.
Table of Contents
- Installation
- Quick Start
- Adding Tasks
- Operating & Running Tasks
- Screenshots
- How It Works
- Contributing
- Roadmap
- License
Installation
# Until v1.0 install the latest pre‑release tag
npm install -g @astrotask/cli@next # or: pnpm / yarn
Upgrading is just as easy:
npm update -g @astrotask/cli
Prerequisites Node 18+, SQLite 3.40+, and
pnpmif you plan to work on the monorepo.
After installation you'll have the astro command in your PATH.
Quick Start
CLI
# 1. Create a workspace
mkdir my-project && cd $_
# 2. Initialise Astrotask (creates ./data/astrotask.db and starter rules)
astro init
# 3. Add tasks and view them
❯ astro task add "Create README"
❯ astro task generate --file docs/task-generation-implementation.md
❯ astro task tree
🌳 Task Tree (All Tasks) (pending and in-progress only)
💡 Use --show-all to include completed and archived tasks
├── ⏳ Create README (AFDR) [pending]
├── ⏳ Task ID System and TaskTree Architecture (BXCQ) [pending]
│ ├── ⏳ Create task creation logic (BXCQ-RKGO) [pending]
│ ├── ⏳ Implement task tree structure (BXCQ-BTTS) [pending]
│ ├── ⏳ Implement task ID validation functions (BXCQ-PDPT) [pending]
│ ├── ⏳ Implement subtask ID generation function (BXCQ-ZGWZ) [pending]
│ ├── ⏳ Implement task ID generation function (BXCQ-QDPA) [pending]
│ └── ⏳ Design database schema for tasks (BXCQ-FUZS) [pending]
# 4. Open the live dashboard (press <c> to toggle completed tasks)
astro dashboard
Programmatic
import { createAstrotask } from "@astrotask/core";
const astrotask = await createAstrotask({
databaseUrl: "./data/astrotask.db",
});
await astrotask.tasks.addTask({
title: "Implement OAuth",
description: "Add Google login",
});
AI / MCP
{
"mcpServers": {
"astrotask-task": {
"command": "npx",
"args": ["@astrotask/mcp"],
"env": {
"DATABASE_URI": "/home/toda/dev/astrotask/data/astrotask.db"
}
}
}
}
OR
astro init
Configure your agent (Cursor, ChatGPT plug‑in, …) with the endpoint and start calling tools such as:
$ Start working on the next task and track your progress in astrotask
{
"name": "getNextTask",
"arguments": { "priority": "high" }
}
Adding Tasks
Manual (CLI)
# Simple capture
astro task add "Write onboarding docs"
# Add under a parent and set priority
astro task add "Design hero section" --parent <parentId> --priority high
Generate from a PRD
Break a spec into structured work:
astro task generate --file ./specs/authentication.prd.md
# Use --dry to preview without touching your DB
Ask an Agent (MCP)
{
"name": "addTasks",
"arguments": {
"tasks": [
{
"title": "Refactor caching layer",
"description": "Move from LRU to ARC",
"priority": "medium"
}
]
}
}
Operating & Running Tasks
| Action | Command |
|---|---|
| See what to do next | astro task next |
| Full list (pending & in‑progress) | astro task list |
| Mark done | astro task done <id> |
| Update fields | astro task update <id> --status done |
| Visualise tree | astro task tree [--root <id>] |
| Validate dependencies | astro dependency validate |
| Interactive dashboard | astro dashboard |
All commands accept --help.
Screenshots

How It Works
@astrotask/coreprovides a type‑safe service layer over the local database.- The CLI/TUI (
astro) talks to the SDK directly. - The MCP server exposes the same operations to AI tools via JSON‑RPC.
- SQLite WAL mode gives safe concurrent access (dashboard + agent + script).
Design principles
- Local‑first – Data should be useful without a network.
- Single source of truth – CLI, SDK & MCP all share the same database file.
- Explicit context – Agents receive structured bundles, never raw SQL.
- Type‑safe all the way – Zod runtime validation mirrors TypeScript types.
Contributing
We ♥ new contributors! See CONTRIBUTING.md and AGENTS.md for philosophy & guidelines.
- Code must pass
pnpm verify(build ➜ type-check ➜ lint ➜ test). - Keep rules/docs in sync with code changes.
- Small PRs > big bang.
Roadmap
| Milestone | Focus |
|---|---|
| v0.2 | ElectricSQL synchronization |
| v0.3 | Web dashboard |
License
Astrotask is released under the MIT License.
常见问题
Astrotask 是什么?
Manage and organize tasks efficiently with AI agent integration. Create, update, query, and track tasks with hierarchical support and real-time feedback. Enhance productivity by leveraging structured task management tools designed for seamless AI interaction.
相关 Skills
技能工坊
by anthropics
覆盖 Skill 从创建到迭代优化全流程:起草能力、补测试提示、跑评测与基准方差分析,并持续改写内容和描述,提升效果与触发准确率。
✎ 技能工坊把技能从创建、迭代到评测串成闭环,方差分析加描述优化,特别适合把触发准确率打磨得更稳。
PPT处理
by anthropics
处理 .pptx 全流程:创建演示文稿、提取和解析幻灯片内容、批量修改现有文件,支持模板套用、合并拆分、备注评论与版式调整。
✎ 涉及PPTX的创建、解析、修改到合并拆分都能一站搞定,连备注、模板和评论也能处理,做演示文稿特别省心。
PDF处理
by anthropics
遇到 PDF 读写、文本表格提取、合并拆分、旋转加水印、表单填写或加解密时直接用它,也能提取图片、生成新 PDF,并把扫描件通过 OCR 变成可搜索文档。
✎ PDF杂活别再来回切工具了,文本表格提取、合并拆分到OCR识别一次搞定,连扫描件也能变可搜索。
相关 MCP Server
文件系统
编辑精选by Anthropic
Filesystem 是 MCP 官方参考服务器,让 LLM 安全读写本地文件系统。
✎ 这个服务器解决了让 Claude 直接操作本地文件的痛点,比如自动整理文档或生成代码文件。适合需要自动化文件处理的开发者,但注意它只是参考实现,生产环境需自行加固安全。
by wonderwhy-er
Desktop Commander 是让 AI 直接执行终端命令、管理文件和进程的 MCP 服务器。
✎ 这工具解决了 AI 无法直接操作本地环境的痛点,适合需要自动化脚本调试或文件批量处理的开发者。它能让你用自然语言指挥终端,但权限控制需谨慎,毕竟让 AI 执行 rm -rf 可不是闹着玩的。
by stickerdaniel
LinkedIn Profile and Job Scraper 是让 Claude 直接抓取 LinkedIn 个人资料、公司信息和职位详情的工具。
✎ 这个服务器解决了招聘和商业调研中手动复制粘贴 LinkedIn 数据的痛点,适合猎头或市场分析师快速获取候选人背景和公司动态。不过,LinkedIn 反爬机制频繁更新,数据稳定性需要持续维护,使用时建议搭配人工验证。