# Veritier
> Agent Trust Gate for AI that can act. Agents should not run a tool call without a gate and evidence.

Veritier sits on a tool call, looks the claim up in a named public record (npm, PyPI, CourtListener, SEC EDGAR, the U.S. Code) or the company policies you bring, and allows or blocks before the agent acts. Every decision leaves a signed record of what was checked and decided. Anyone can open that record at a public URL. You can revoke it later. It is not a certificate of truth.

An Agent Trust Gate is not a draft checker, not Document Integrity, and not Proof of Reality. Checking drafts for invented claims and scanning files are supporting work. They do not authorize an action.

The gate decides whether an agent gets to act. A draft check reviews writing against live evidence. A draft check never blocks the action by itself.

## Canonical pages

- [Home](https://veritier.ai/): Agent Trust Gate landing, pricing, and FAQ
- [Docs](https://veritier.ai/docs): REST, MCP, signed decision records, and how to wire the gate
- [Engine](https://veritier.ai/engine): Try the gate in the browser
- [skill.md](https://veritier.ai/skill.md): Agent skill. Point agents here.
- [FAQ](https://veritier.ai/#faq): What an Agent Trust Gate is, how agents gate a tool call, gate vs draft check, signed decision record

## Optional

- [Open-source examples](https://github.com/SalomonHenao/Veritier-Fact-Checking)
- [Live OpenAPI](https://api.veritier.ai/openapi.json)