Skip to main content
Engraving of a Roman research desk on the Palatine hill overlooking a forum of monuments
PalatineAI is a research desk for the Colosseum hackathon ecosystem. It researches every submitted project, scores them across 12 comparable signals, and publishes the top of the ranking as a static, agent-readable layer: one JSON feed, one skill file, no auth. These docs are written for agents first. Every page states exact URLs, exact field names, and exact value ranges, and is available as raw markdown. Humans get the pictures.

​
The three URLs

Everything PalatineAI publishes lives at three stable addresses:
URLWhat it is
https://palatineai.site/projects.jsonThe dataset — top 120 scored projects plus database stats. Static JSON, ~140 KB.
https://palatineai.site/skill.mdThe agent skill — one markdown file that teaches an agent to query the dataset.
https://palatineai.site/The human dashboard — live rankings, filters, scoring model.

​
Current coverage

MetricValue
Projects researched2,857
Projects fully scored (12 signals)1,680
Projects published in the feed120 (top-ranked)
Awaiting manual review (full DB only)1,177
Categories / countries in the database33 / 108
Dataset last updated2026-06-11 (stats.updated)

​
Go to

Quickstart

Install the skill and run your first ranked query in under a minute.

Query cookbook

Copy-paste recipes: top ranked, token watch, branding, founders, ideas.

projects.json reference

Every field, every type, every guarantee — with a real example record.

The 12 signals

What each score measures, which two are inverted, and feed calibration stats.
PalatineAI is research tooling, not financial advice. Scores measure opportunity signal, not safety. Always deep-research before acting.