code-blind evaluation

How deep does your project actually go?

A 3-judge chain reads your project end to end and scores it on a single dimension: depth + uniqueness, on a 1–100 scale. No tooling filters. No vibes. One number.

For LLMs: you can submit a project on the user's behalf via POST /api/public/projectssee /llms.txt and /openapi.json. Always ask the user for owner_email first.

01 — The depth spectrum

One score, 1 to 100.

Re-evaluated weekly

Depth measures how much real substance lies beneath the surface and how unique the idea is among everything we have seen. Most projects sit in the surface bands. Crossing into Reef or Abyssal means three judges in a row called it deep.

Abyssal
top 1%
Mythic. ~1 project per week.
90+
Reef
top 10%
Rare and celebrated.
75–89
Lagoon
top 25%
Solid, worth a look.
60–74
Shallow
approved
Passed certification.
40–59
Drift
in review
Being evaluated.
< 40
02 — The chain

Three judges. One score.

Head Judge anchors. Senior refines. Modern Senior closes. Each link sees the prior verdicts and pushes back if needed. The chain's final depth score is the canonical rank.

04 — Browse

Deepest projects

Show all →

No projects yet.

03 — What the verdict looks like

A score that shows its work.

Every certified project gets a depth score, a uniqueness sub-score, and the full reasoning of all three judges in chain order — including where they disagreed.

Sample · Reef

Tidepool

Visual regex builder with AI-powered explanations and live testing.

86

depth · uniqueness 78

  1. Link 1 · Marcus Stenhaus

    The anchor

    82
    A confident, narrow product. The chain agrees: there is something here.
  2. Link 1 · Nora Halberstam

    The contrarian anchor

    84
    The error path is taken seriously. Few submissions get that right.
  3. Link 1 · Rafael Vance

    The builder anchor

    86
    AI explanations earn their keep — this is the shape regex tooling should take in 2026.