AI News
商业产业 Hacker News AI 2026-09-24T17:21 正文 ok 785 words

Show HN：Whiteboard（YC W26）——用于深度软件设计的开源 IDE

英文标题：🔶 Show HN: Whiteboard (YC W26) – An open-source IDE for thoughtful software design

Whiteboard 是一款开源桌面应用，让开发者与编码 Agent 在同一工作区内协作进行软件架构设计和代码审查。它可连接 Claude Code、Codex 等工具，并通过 SDK 让 Agent 在应用内画布上绘制流程图、关系图和工作轨迹说明。产品提供基于 Rust 的语义化、AST 感知 Diff 查看器，以及 WASM 插件系统，旨在降低传统代码 Diff 的噪声。Whiteboard 采用 MIT 许可证、面向本地代码检出运行；团队托管产品正在规划中，但项目承诺始终支持自托管。
打开原文相关 4 · 机会 2
media12.7 KB media528.9 KB media46.9 KB

中文

Whiteboard 是一款开源桌面应用，人类和 Agent 可以在同一个工作区中共同设计软件架构。

Whiteboard 可接入你已经在使用的工具，例如 Claude Code、Codex 等，并为你的 Agent 提供 SDK，使其能够在应用内画布上绘图来说明自己的工作。

一名 Agent 在 Whiteboard 中绘制流程图，旁边是其所描述的代码

以下是一段时长 1 分钟的演示视频，介绍更多内容：视频/嵌入内容：Whiteboard 1 分钟演示视频

  1. 下载 Whiteboard并打开应用。
  2. 在欢迎界面连接 Claude Code、Codex 或其他编码 Agent。
  3. 让 Agent 将你当前分支与最新的 main 分支进行对比审查，并在 Whiteboard 中打开结果。

根据我们的经验，Whiteboard 最适合搭配 GPT-6 Sol 和 Claude Opus 5.5 等模型使用，它们在智能水平、成本和速度之间具备较好的平衡。

下面是一些有效使用 Whiteboard 的提示词示例。我们正努力确保正确的选择会默认内置于系统提示词中——这也是该系统开源的部分原因！——但在此之前：

嘿，这组提交的设置目标是让我能通过一个 [API] 完成 [目标]

我希望看到：

  • 建议的 API
  • 示例
  • 这样做的动机（如果你能从上下文或仓库中获取）

然后我们可以进一步讨论实现细节，并解释它是如何工作的。

为新增遥测功能进行变更

你能否向我解释最新 PostHog PR 中的遥测变更吗：https://github.com/devdotfast/whiteboard/commit/4837e107946e27ebad50c282eb0f2585210d2a35 -- 我们在追踪什么？如何基于这些数据构建好的仪表盘或产品漏斗？对于卡死、错误、崩溃等情况，我们该怎么处理？请使用 Whiteboard。

如果你看到任何不喜欢的内容，可以在剪贴板中选中它并交给 Agent；它就能重新在 Whiteboard 上绘制内容，以满足你的需求。

从图表走向代码

纯 HTML 工具无法轻松地将规范或图表与代码关联起来；这尤其棘手，因为许多权衡往往要在第一轮实现之后才会显现。在 Whiteboard 中，当你点击时序图、实体关系图，或 Agent 工作轨迹中的引用等可视化内容时，可以直接跳转至其底层代码。在浏览代码时，你可以开箱即用地获得来自 VSCode 的快捷键和 LSP 支持。

原始 Diff 视图可能非常嘈杂，因此我们用 Rust 编写了一个具备语义理解和 AST 感知能力的 Diff 查看器，让你只查看与自己相关的代码变更。我们设置了一些合理的默认值：新增加的大型函数会被总结为伪代码，单元测试和文档改动等内容则会被折叠或隐藏。这些行为都可以通过基于 WASM 的插件系统进行自定义。

我们发现，很难厘清 Agent 自主做出了哪些决策，以及这些决策会如何影响一次变更。因此，我们构建了让 Agent 在 Whiteboard 上查询并链接其自身工作轨迹的工具，使你能够可视化自己设定的需求，理解这些需求如何被实现，并了解 Agent 自主作出了哪些决策。

开源，运行在你的机器上

Whiteboard 采用 MIT 许可证，并针对你的本地代码检出运行。我们计划推出面向团队的托管产品，但一切都将始终保持可自托管。

  • 目前无法在 Whiteboard 中编辑文件。如果这是你希望使用的功能，请提交 issue！

  • 尚未很好地支持在一次审查中跨多个仓库处理和浏览文件。

  • 虽然你可以通过分享按钮在多台机器之间共享审查结果，但审查被共享后所作的更新不会显示给其他人。你需要重新分享该审查。

欢迎贡献和反馈。

请阅读 CONTRIBUTING.md 了解环境设置和 Pull Request 流程，并遵循行为准则。请按照 SECURITY.md 中的说明报告漏洞。有问题？请前往 Discord 提问。

Whiteboard 针对本地代码检出运行。匿名遥测数据不会包含你的代码、Diff、Whiteboard 文本、提示词或模型输出。你可以阅读隐私概览，查看完整的遥测参考文档，或随时关闭遥测功能。

Whiteboard 基于 MIT License 发布。随附的 Code - OSS 分支保留 Microsoft 的 MIT 许可证及第三方声明；详见 apps/review-desktop/LICENSE 和 apps/review-desktop/UPSTREAM。

如今，所有人都在使用专用的 Agent TUI 和桌面应用，我们现在只会用文本编辑器逐行审查 Diff。因此我们想，何不做一个专门用于代码审查的文本编辑器？既然如此，不如以目前最成功的开源编辑器为基础开始。

与通过维护补丁来实现的其他分支不同，我们直接随附 Code OSS，因为编码 Agent 很难处理补丁，而且原版 VS Code 中有许多我们不需要的内容（例如，如今代码库中约 45% 都是 Copilot 😬）。

我们会定期监测上游 Code OSS，并在安全补丁和功能补丁发布后进行合并。

English / 原文

Whiteboard is an open-source desktop app where humans and agents can architect software together in a common workspace.

Whiteboard plugs into the tools you already use - e.g. Claude Code, Codex, etc. – and gives your agent an SDK to draw on an in-app canvas to describe its work.

An agent draws a flow diagram on a Whiteboard next to the code it describes

Here’s a 1 min demo video explaining more: https://www.youtube.com/watch?v=ChPn3ftULWE

  1. Download Whiteboard and open the app.
  2. Connect Claude Code, Codex, or another coding agent from the welcome screen.
  3. Ask your agent to review your current branch against up-to-date main and open the result in Whiteboard.

In our experience, Whiteboard works best with models like GPT-6 Sol and Claude Opus 5.5 for their intelligence, cost, and speed tradeoff.

Here are a few example prompts of how to use Whiteboard effectively. We are working hard to make sure the right choices are baked in by default to the system prompt - part of why this system is open source! - but in the meantime:

hey, this stack of commits is set up so i can get an [api] to do [objective]

i'd like to see:

  • proposed api
  • examples
  • motivations for this (if available to you in context/in the repo)

and then we can dive into implementation + explaining how things worked.

For a change to add telemetry:

cna you explain to me the telemetry changes form the newest posthog pr https://github.com/devdotfast/whiteboard/commit/4837e107946e27ebad50c282eb0f2585210d2a35 -- what are we tracking, how can we build good dashboards or product waterfalls from it? what do we do for hangs, errors, crashes etc... use whiteboard

If you see anything you don't like, highlight it in your clipboard and give it your agent, and it can re-draw on the Whiteboard to suit your needs!

Diagrams that lead to code

Pure HTML tools didn’t provide easy affordances to connect a spec or diagram to code; this is especially tricky since tradeoffs are often only discovered after a first pass at implementation. In Whiteboard, when you click on visualizations like a sequence diagram, an entity relationship diagram, or a quote from the agent’s trace, you can jump to the underlying code directly. When navigating code, you get keybindings and LSP support from VSCode out of the box.

Raw diff views can be very noisy, so we wrote a semantic, AST-aware diff viewer in Rust so you can only view the code changes which are relevant to you. We’ve set up some sane defaults: large added functions are summarized as pseudocode, and things like unit tests and documentation changes are collapsed / hidden. This is all customizable with a WASM-based plugin system.

We found it difficult to reason about what set of decisions our agents made autonomously & how that impacts a change. So we built tools for agents to query and link their own traces on the Whiteboard, so you can visualize the requirements that you set, understand how they were implemented, and understand what decisions the agent made autonomously.

Open source, on your machine

Whiteboard is MIT-licensed and runs against your local checkouts. A hosted product for teams is planned, and everything will always remain self-hostable.

  • You cannot currently edit files in Whiteboard. If this is something that you find yourself wanting to do, please file an issue!

  • Working and browsing files across multiple repos in a single review isn't well supported.

  • While you can share reviews between machines with the share button, updates made after a review is shared don't appear for others. You would need to re-share the review.

Contributions and feedback are welcome.

Read CONTRIBUTING.md for setup and the pull request workflow, and follow the Code of Conduct. Report vulnerabilities as described in SECURITY.md. Questions? Ask on Discord.

Whiteboard runs against local checkouts. Anonymous telemetry does not include your code, diffs, Whiteboard text, prompts, or model output. Read the privacy overview, inspect the complete telemetry reference, or turn telemetry off at any time.

Whiteboard is available under the MIT License. The vendored Code - OSS fork retains Microsoft's MIT license and third-party notices; see apps/review-desktop/LICENSE and apps/review-desktop/UPSTREAM.

With everyone using dedicated agent TUIs and desktop apps, we only use our text editors for reviewing line-by-line diffs now, so we figured why not have a text editor meant for reviewing code. In that case, might as well start off with the most successful open source editor out there as a baseline.

We vendor Code OSS unlike other forks that maintain patches because coding agents have a hard time with patches and there's a lot of stuff from stock VS Code (i.e., ~45% of the codebase is Copilot these days 😬) that we don't need.

We regularly monitor upstream Code OSS and merge in security/feature patches as they come in.