Visualize what an AI coding assistant changed instead of reading thousands of generated lines.
Explore a project's Epics, Features, and Steps in an interactive diagram.
Watch the flow graph update automatically as an AI agent writes features.json.
Keep a shared, provider-agnostic map of a large AI-assisted codebase.
| kaka-cheaper/codesee | agentfm-ai/agent-fm | ame-x/auto-ppt | |
|---|---|---|---|
| Stars | 21 | 21 | 21 |
| Language | TypeScript | TypeScript | TypeScript |
| Setup difficulty | easy | easy | moderate |
| Complexity | 2/5 | 2/5 | 2/5 |
| Audience | developer | developer | vibe coder |
Figures from each repo's GitHub metadata at analysis time.
Requires an AI coding assistant that writes a features.json file in the expected format.
CodeSee is a developer tool that turns your software project into a visual, human-readable story, a "semantic flow graph", instead of the usual code maps that show which function calls which other function. The idea is that when an AI coding assistant writes thousands of lines of code for you, it also writes a special file called features.json that describes what each feature does in plain steps, like "prep → crack eggs → heat oil → stir-fry → season → plate." CodeSee then renders those steps as an interactive diagram you can explore in a browser. The graph has three levels you can drill into: Epics (big goals), Features (individual capabilities), and Steps (the detailed flow within a feature, including branches for errors or async operations). The viewer polls the features.json file every three seconds, so the diagram updates in near real-time as the AI works. You can drag nodes around, zoom in and out, and the layout is saved automatically between sessions. You would use CodeSee when you are collaborating with an AI coding tool, like Cursor, Claude Code, or GitHub Copilot, on a project large enough that you can no longer keep the whole thing in your head. Instead of reading thousands of lines of generated code to understand what changed, you look at the graph. The entire system is built on a plain JSON file, so it works with any AI provider and requires no proprietary lock-in. It is built in TypeScript and uses the React Flow library for the interactive canvas. The full README is longer than what was provided.
A TypeScript tool that turns an AI coding assistant's generated feature descriptions into an interactive, real-time visual flow diagram.
Mainly TypeScript. The stack also includes TypeScript, React Flow, JSON.
Setup difficulty is rated easy, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.