# PalatineAI

> PalatineAI documentation — Colosseum hackathon intelligence packaged as an agent-readable skill and dataset.

## Docs

- [Agent access](https://docs.palatineai.site/agents/access.md): Machine-readable surfaces, freshness checks, and honest transport details
- [Query cookbook](https://docs.palatineai.site/agents/queries.md): Copy-paste recipes against projects.json — no API, just one static file
- [Install the skill](https://docs.palatineai.site/agents/skill.md): One markdown file that teaches any agent to query PalatineAI
- [Evaluation workflow](https://docs.palatineai.site/agents/workflow.md): From feed to decision: TRACK, CONTACT, BUY/WATCH, or SKIP
- [Changelog](https://docs.palatineai.site/changelog.md): Dataset refreshes and artifact versions, newest first
- [Methodology & coverage](https://docs.palatineai.site/dataset/methodology.md): How 2,857 projects become 120 published rows — and what never makes it in
- [projects.json reference](https://docs.palatineai.site/dataset/projects-json.md): Every field, every type, every guarantee in the published feed
- [The 12 signals](https://docs.palatineai.site/dataset/signals.md): What each score measures, which two are inverted, and how the feed is calibrated
- [PalatineAI](https://docs.palatineai.site/index.md): The evaluation layer for new opportunities — Colosseum hackathon intelligence, packaged for agents
- [Quickstart](https://docs.palatineai.site/quickstart.md): From zero to ranked Colosseum intelligence in under a minute