Skip to main content
Engraving of Hermes handing a sealed scroll to a mechanical owl
Two artifacts do all the work: skill.md (teaches your agent the dataset) and projects.json (the dataset itself). Both are static files — no API keys, no signup, no rate-limit dance.
1

Install the skill

For Claude Code, one command:
Install for Claude Code
mkdir -p ~/.claude/skills/palatineai && curl -fsSL https://palatineai.site/skill.md -o ~/.claude/skills/palatineai/SKILL.md
Any other agent framework: fetch https://palatineai.site/skill.md and load it as context or a tool description. Details and variants in Install the skill.
2

Run your first query

Top 10 by total opportunity score
curl -fsSL https://palatineai.site/projects.json | jq '.projects[:10] | map({rank, project, category, total})'
First rows today (dataset 2026-06-11):
[
  { "rank": 1, "project": "QSTN",    "category": "Consumer Apps",        "total": 83.75 },
  { "rank": 2, "project": "Encrypt", "category": "ZK / Crypto Research", "total": 83.5  },
  { "rank": 3, "project": "Lucilla", "category": "Consumer Apps",        "total": 83.5  }
]
3

Ask a sharper question

The feed answers more than “who’s first.” Strongest projects with no token yet:
Token-watch shortlist
curl -fsSL https://palatineai.site/projects.json | jq '[.projects[] | select(.scores.noToken >= 8) | {project, momentum: (.scores.socialActivity + .scores.branding + .scores.ideaQuality), x}] | sort_by(-.momentum)[:10]'
The full recipe set — categories, founders, branding, league tables — is in the query cookbook.

​
Where next