I build tools for people making software with coding agents, and websites they can make their own.
Website templates with their own visual direction. Explore the live sites and choose a starting point for your project.
Find the instructions you keep repeating to your coding agents. Emulo reads your session history, shows dated evidence and helps turn it into an editable you.md profile.
Start with a report of your own history:
uvx emulo --coachHow profiles work and which agents can load them.
Inspect an AI-built app before launch. The local Studio brings together provider context, release diffs and checks for gaps around authentication, billing and database access.
Run this from your project:
npx -y viberavenRepository checks are a starting point. They do not establish that your deployed app is secure or that its payment flow works.
Public repository and setup instructions.
I ran a pre-registered, 60-run experiment on Emulo: no profile, my real profile, and a length-matched fictional profile.
None of my three predictions met the pass rule against both controls. Saturation changed versus no profile, but that result did not separate my profile from the placebo.
I published the result and its limitations. Extracting a recognizable preference and proving it improves an agent's work are separate questions.
If you tried one of these tools, tell me what you were trying to do and where it helped or got stuck. If you work on coding agents or their evaluations, I would also like to compare notes on testing memory and instruction following.