Agent View
The site as an AI agent gets it, beside the site as a person sees it. What it can read, what it has to cut away, and what it never sees.
ProductPreset Cloud Fully-managed, cloud-hosted…Managed Private Cloud Preset with additi…Preset Certified Superset Deploy QA-appr…Preset Embedded Dashboards Interactive a…Preset API Managing your Preset workspac…AIPreset Chatbot Conversational analytics,…
SUPERSET DEVELOPERS
Introducing Claudette CLI: Supercharge Superset Development
Maxime Beauchemin, Evan Rusackas
The rest of the page is in full screen.
SUPERSET DEVELOPERS # Introducing Claudette CLI: Supercharge Superset Development Maxime Beauchemin, Evan Rusackas August 28, 2025 3 min read 481 words **TL;DR**: Tired of juggling multiple Superset environments and waiting for containers to restart every time you switch branches? Meet Claudette CLI, a tool that lets you run multiple isolated Superset environments simultaneously, built for the AI-assisted development era. ## Overview: See Claudette In Action ## The Problem: Environment Hell If you've ever worked on Apache Superset development, you know the pain. Switching between feature branches means: - Rebuilding Docker containers (again) - Reinstalling npm dependencies (again) - Dealing with Python virtual environment conflicts - Port collisions when trying to run multiple instances - Lost context and broken workflows In today's AI-assisted development world, where Claude Code and other AI tools can generate code rapidly, these environment bottlenecks have become the new limiting factor. You're no longer waiting for code to be written; you're waiting for environments to spin up. ## The Solution: Claudette CLI Claudette CLI transforms how you work with Superset by using Git worktrees to create completely isolated development environments for each project or branch. You end up with multiple Superset installations that share the same Git history but run independently. ### Key Features **🌳 Git Worktree Magic**: Each project gets its own directory with isolated Python virtual environments and node_modules, while sharing the underlying Git repository. **🐳 Docker Integration**: Run multiple Superset instances on different ports simultaneously—no more container conflicts. **⚡ Lightning Setup**: New projects spin up in seconds with automatic dependency installation using `uv` for Python packages. **🧊 Smart Resource Management**: Freeze inactive projects to reclaim ~3GB of disk space per environment, then thaw them instantly when needed. **🤖 Claude Code Ready**: Built-in integration with Claude Code for seamless AI-assisted development workflows. **📝 Project Documentation**: Persistent PROJECT.md files that survive branch switching and provide context to AI assistants. ## Quick Start Getting started takes less than a minute: ``` # Install claudette pip install git+https://github.com/mistercrunch/claudette-cli.git # Initialize (clones Superset and sets up base config) claudette init # Create your first isolated project clo add my-feature # Drop into a Docker container and start coding clo shell ``` Pro tip: Use `clo` as the shorter alias for all commands! ## Real-World Impact The benefits become immediately apparent: - **Parallel Development**: Work on multiple features simultaneously without environment conflicts - **Zero Context Switching Cost**: Each environment maintains its state—databases, configurations, and dependencies - **AI-Optimized Workflow**: Spend time coding with AI assistance instead of waiting for environments - **Resource Efficient**: Freeze unused projects to keep your machine lean ## Perfect For - **Superset Contributors**: Manage multiple PRs effortlessly - **Data Teams**: Test features across different branches simultaneously - **AI-Assisted Developers**: Eliminate environment setup bottlenecks - **Anyone juggling multiple Superset projects** ## Commands You'll Love ``` clo add feature-x # Create new isolated environment clo shell # Enter Docker container clo docker up # Start services clo claude code # Launch Claude with project context clo pytest tests/ # Run tests in isolation clo freeze old-feature # Save 3GB+ disk space clo list # See all your projects ``` ## Get Started Today Ready to transform your Superset development workflow? **➡️ [Get Claudette CLI on GitHub](https://github.com/mistercrunch/claudette-cli)** The tool is open source and actively maintained. Contributions, feedback, and feature requests are welcome! All the features of Superset (and more) without the hassles of scaling/managing/upgrading, with plenty of time to grow into it. Free for five users, forever. --- **Related Reading** - [How Micro-Corrections Drive AI Success](https://preset.io/blog/how-micro-corrections-drive-ai/): Why intuition and small nudges matter more than perfect prompts - [Adopting Claude Code: Riding the Software Economics Singularity](https://preset.io/blog/adopting-claude-code-riding-the-software-economics-singularity/): The broader transformation in software development economics - [How to Contribute to Superset](https://preset.io/blog/how-to-contribute-to-superset/): Getting started as a Superset contributor ## Let's get visual. Try Preset today. [Start for free](https://manage.app.preset.io/starter-registration/) #### Subscribe to our blog updates Receive a weekly digest of new blog posts
An agent gets this page whole: its own text is in the HTML the server sends.
The figures
100 pages, set against the 167,893 sites with figures.
- Readablehigher is better
- 94%94 of 100 pages
- All 167,893 sitesBetter than 34%
Pages whose own text is in the HTML the server sends, enough of it for their kind.
- Cleanhigher is better
- 91%middle page
- All 167,893 sitesBetter than 87%
Of the words before script, the share that are the page and not menus, footer and forms.
- Costshown, never scored
- 309,412tokens
- About 3,094 a page
To read every page once. A long page that answers well is not worse for it. Tokens are estimated at about 4 characters each.
How its pages spread on Clean
100 pages measured.
Against its rivals
The sites nearest it by what they do, with figures. Close means similar work, not better.
| Readinghigher is better | Costnot scored | ||
|---|---|---|---|
| Site | Readable | Clean | Tokens a page |
| preset.ioThis site, 100 pages | 94% | 91% | 3,094 |
| itpathsolutions.com100 pages | 98% | 96% | 6,200 |
| formidable.com100 pages | 100% | 95% | 2,544 |
| expanso.io100 pages | 100% | 93% | 3,248 |
Fourth of 4 on Readable, fourth on Clean.
AI crawler access
What robots.txt says to each, as read on October 24, 2026. 8 of 8 may read the site.
| Crawler | robots.txt says | Used for | The line that decides |
|---|---|---|---|
| GPTBotOpenAI | Not named | Training models | Falls to User-agent: * |
| OAI-SearchBotOpenAI | Not named | ChatGPT search | Falls to User-agent: * |
| ChatGPT-UserOpenAI | Not named | A page a person asks about | Falls to User-agent: * |
| ClaudeBotAnthropic | Not named | Training models | Falls to User-agent: * |
| Claude-SearchBotAnthropic | Not named | Claude’s search | Falls to User-agent: * |
| PerplexityBotPerplexity | Not named | Perplexity search | Falls to User-agent: * |
| Google-ExtendedGoogle | Not named | Training Gemini | Falls to User-agent: * |
| bingbotMicrosoft | Not named | Bing and Copilot | Falls to User-agent: * |
Every AI crawler mttr checks may read it. Not named means the rules for every crawler apply.
Every page
Worst first. Each opens in Agent View at the top.
6 pages of 100 lose something. Pick a loss to see what to do about it.Open the action plan
Show where it stands
A card, a badge and a line, each with the figures.

A badge for the site
<a href="https://mttr.si/agent-view/preset.io"><img src="https://mttr.si/agent-view/preset.io/badge.svg" alt="Agent View: 94% readable, 91% clean" height="20"></a>
The figures in one line
preset.io in Agent View: 94% of pages readable, better than 34% of sites; 91% clean; 309,412 tokens to read all 100 pages. mttr.si/agent-view/preset.io
What to do next
Fix the pages, let an agent do it, and read the site again to see the figures move.
Fix what agents miss
- Render the rest of 6 pages on the serverReadable
Connect your agent
Claude Code, Codex or Cursor reads the site’s fixes over MCP, fixes a page in your code, has mttr read it again, and checks the fix landed.
https://mttr.si/mcpSet up an agent
Claim the site
Prove it is yours with a DNS record or a file. Your agent can then reach its reads and jobs, and set each change to wait for your approval.
Claim the siteDrawn from the HTML mttr fetched on its read of October 24, 2026, not a screenshot, and only from the 100 pages read, where the read stopped at its page limit. An assistant’s own reader may differ. Tokens are estimated at about 4 characters each. A site an agent reads well is not a site an assistant cites.