Whiteboard

Product Introduction

1. Definition: Whiteboard is an open-source, desktop-first software development canvas designed for architectural code review. It falls into the technical categories of developer tooling, code review platforms, and AI-assisted development environments.
2. Core Value Proposition: Whiteboard exists to elevate code review from line-by-line commentary to a holistic, architectural critique. Its core value is providing a unified interface to visualize diffs, audit AI agent execution trajectories, and interact with code contextually, thereby enabling more thoughtful and efficient software development workflows.

Main Features

1. Semantic, AST-Powered Diffs: Whiteboard does not display raw text diffs. Instead, it uses its underlying Rust library, diffr, to perform Abstract Syntax Tree (AST)-based differencing. This allows for intelligent features like folding long import sections, summarizing complex algorithms into readable pseudocode while keeping the full source code expandable, and consistently showing function signatures for context.
2. Interactive Architecture Visualizations: The platform transforms code changes into interactive diagrams. Developers can visualize the impact of a pull request through automatically generated sequence diagrams, database schema changes, or call trees. This feature moves review focus from syntax to system behavior and data flow.
3. AI Agent Trajectory Auditing: As AI coding assistants become prevalent, Whiteboard addresses the "black box" problem. It allows developers to trace and audit an AI agent's complete decision-making trajectory within a task. Reviewers can see the agent's initial instructions, intermediate steps, code edits, and final output, enabling validation of both the process and the final implementation.
4. Context-Aware Diff Search API: A proposed API feature demonstrates advanced code search integration. It allows parsing search results (e.g., from ripgrep --json), hydrating them with full file context from a pinned Git commit, and post-processing selections with existing plugins. This enables workflows where search results become interactive, reviewable artifacts.

Problems Solved

1. Pain Point: Traditional code review in pull requests is often myopic, focusing on syntax errors and style in isolation, missing architectural consequences and breaking the developer's flow.
2. Target Audience: Senior Developers, Engineering Leads, and Architects who need to ensure system integrity; Developers and Teams heavily utilizing AI coding assistants (like GitHub Copilot, Cursor, or Claude Code) who need to validate AI-generated work; Open-source maintainers reviewing complex contributions.
3. Use Cases: Reviewing a database schema migration and visualizing its impact on existing queries; Auditing a series of commits from an AI agent to ensure it correctly implemented a specified refactor; Onboarding a new team member by walking through architectural decisions visualized from recent diffs; Conducting a security audit by tracing data flow paths through sequence diagrams.

Unique Advantages

1. Strengths & Limitations (Pros & Cons):

   • Pros: Unifies code review, visualization, and AI audit into a single, dedicated canvas. AST-based diffs are fundamentally more semantic and readable than text-based ones. The focus on "trajectories" is forward-thinking for the AI-augmented development era. Being open-source allows for customization and self-hosting.
   • Cons: As a desktop application, it requires installation and operates locally, which may not integrate as seamlessly into cloud-based CI/CD pipelines as web-native tools. The visual and architectural review focus may be overkill for simple, trivial changes. The ecosystem and plugin library is nascent compared to established review tools.

2. Key Alternatives & Differentiation:

   • GitHub Pull Requests / GitLab Merge Requests: The standard. Whiteboard differentiates by moving beyond inline comments on text to interactive visualizations and architectural context. It is a complementary deep-review tool, not a direct replacement for the social approval workflow of PRs/MRs.
   • Reviewable, Phabricator: These are also dedicated code review tools. Whiteboard's key differentiation is its native, first-class support for AI trajectory auditing and its strong emphasis on generating visual diagrams (sequence, DB) directly from code changes, which are not core features of these alternatives.
   • Miro or Excalidraw (for diagrams): While used for architectural diagrams, they are general-purpose and require manual, error-prone drawing. Whiteboard automatically generates accurate, interactive diagrams directly from the code diff, ensuring the visualization is always synchronized with the actual implementation.

Frequently Asked Questions (FAQ)

1. Is Whiteboard a replacement for GitHub Pull Requests? No, Whiteboard is a complementary tool for deep, architectural review and AI audit. It is designed for the thoughtful analysis phase, while platforms like GitHub handle the formal approval, merging, and integration workflow. They are best used together.
2. How does Whiteboard visualize code changes? Whiteboard uses an in-house, Rust-based AST-diffing library (diffr) to parse code structure. It then applies rules and plugins to render diffs semantically (e.g., as pseudocode) and can generate interactive diagrams like sequence flows or database schemas by analyzing the changed code paths and data models.
3. Can I use Whiteboard to review AI-generated code? Yes, this is a primary use case. The "Audit agent trajectories" feature allows you to import or trace the complete step-by-step output of an AI coding agent, review its decision logic, and verify that the final code changes correctly implement the intended task.
4. Is Whiteboard free and open-source? Yes, the core Whiteboard application is open-source (available on GitHub under the devdotfast organization) and free to use. The team has indicated a "Coming Soon" hosted, team-based version, which may involve a commercial offering.
5. What programming languages does Whiteboard support? The provided examples focus on TypeScript/JavaScript. Support for other languages depends on the underlying diffr AST library's capabilities. The open-source nature means community can contribute parsers for additional languages.