Native powers.
One team.
Each seat keeps the tools that make it good. The engine makes them one team.
Nova shares out seven tasks across six seats and four model families, with live status.
It conducts. It doesn’t get the final say.
You name it. It plans, assigns, watches budget and evidence, and tells you when it’s unsure. It can recommend anything. It can approve nothing.
Personality makes it easy to work with. Separating it from authority makes it safe.
Different minds. Different tools. One context.
A seat is a role. The model behind it is chosen by results, not brand.
Scout
Finds, compares and cites sources. Works the web in a safe browser.
Forge
Writes and tests code in its own coding environment and terminal.
Iris
Designs screens and visuals, and keeps them consistent.
Lens
Reviews work it didn’t make, against the plan and PROJECT.md.
Sentinel
The built-in adversary. Different model family, real evidence, no authority.
Hush and Moss
Local models on your computer for anything that shouldn’t leave it.
Quill
Turns research and decisions into clear documents, posts and emails.
Add your own
Bring any model you have a key for, or any local model, as a new seat.
“A seat is a role. The model behind it earns its place.”
Seats are chosen by reviewed results, availability and cost, not brand.
They check each other’s work
Challenges point at the exact line and the fix expected. Nothing moves until a seat that didn’t write it agrees.
A real review: Lens challenges Forge, Forge fixes it, Sentinel verifies the evidence.
Earned, not claimed.
Grades count only work that passed independent review. When a seat falls back to a backup model, it says so.
Names and grades shown are examples.
Save your seat
at the table.
We’re inviting people in waves. Everyone who joins with your link moves you up.
Share your link. Each person who joins with it moves you up 50 places.
By joining you agree to the terms and privacy notice.
