rolldeep

rolldeep

Enables AI agents to operate a local study lab by rolling one-time topics, choosing a card, and completing it with proof, while managing syllabi and progress locally.

Category
访问服务器

README

ROLLDEEP

You don't have a syllabus problem. You have a choosing problem.

Free local study lab. Hit ROLL. Get one topic, one curiosity question, and a 25–60 minute window. Study however you want. Record a video so you actually did it. That topic never comes back.

No account. No cloud. MIT. Runs on your machine like Jupyter.

ROLLDEEP — hit ROLL, get one topic, do the work

Watch the 20-second demo

Why this exists

Study apps sell choice: dashboards, playlists, “continue where you left off.” That is how you open Notion, pick the easy chapter, or pick nothing.

ROLLDEEP sells the opposite. The ROLL button is the product. You can reroll a few times. Then you TAKE one. Then you finish.

Who it is for

  • Students with a real syllabus who keep planning instead of starting
  • Anyone who wants one locked-in deep-work block instead of a to-do list
  • People who need proof they studied — a video is required to complete

Start in one command

git clone https://github.com/dev-the-dev-while-deving/rolldeep-lab.git
cd rolldeep-lab
npm install
npm rebuild better-sqlite3
npm link
rolldeep

That opens the lab at http://127.0.0.1:3210. Same idea as jupyter lab. Aliases: rolldeep lab, rolldeep notebook.

Without a global link: npx rolldeep.

Runtime data stays in ~/.rolldeep/ (override with ROLLDEEP_HOME). The topic pool is files in this repo: content/units/*.json.

Ships with a CS2000 Design and Analysis of Algorithms pool. Each content/units/*.json file is its own syllabus. Switch with the dropdown in the lab — rolls, the pool, and completions stay separate per course.

Drop another syllabus in content/syllabi/ and have an agent (or a human) write a new units file. Then pick it from the dropdown.

How a session works

  1. Pick a syllabus in the dropdown. Hit ROLL — up to 5 times on that course. You see the current card plus 3 previous. TAKE one.
  2. Study however you want. Make a short video so you actually did the work.
  3. Complete with that proof URL. The topic is gone. Forever.

Each card shows topic, then question, hardness stars (1–5), and a time window. There is no study how-to in the UI. You already know how to study. You needed something that would not let you stall.

The rules

  • Up to 5 rolls per session; only current plus 3 previous stay visible
  • TAKE one, then finish with proof
  • Completed topics are excluded forever on that syllabus
  • Multiple syllabi; dropdown (or rolldeep use) switches course without mixing progress
  • Topic name and curiosity question are both shown
  • Local only — your work never leaves the machine

Free, on purpose

ROLLDEEP is free software (MIT). No signup, no telemetry, no “pro” gate on the ROLL button. Fork it, load your syllabus, keep the streak.

If you want more units, add JSON. If you want an agent to operate it, the harness is first-class.

For any AI model

Same lab. Three ways in:

Path How
Files Edit content/units/*.json, then rolldeep sync
CLI rolldeep syllabi / use / status / roll / choose / complete
MCP rolldeep mcp — or add the server from .mcp.json
rolldeep syllabi
rolldeep use --title "Course name"
rolldeep status
rolldeep roll
rolldeep choose --id <id>
rolldeep complete --id <id> --url https://...
rolldeep sync

Grok:

grok mcp add rolldeep -- npx tsx mcp/server.ts

Claude Code / Cursor: this repo already has .mcp.json.

Read SKILL.md and AGENTS.md. Do not mint dummy “What is X?” questions. Author units into the JSON files.

License

MIT

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