# Headron

> Headron is the trust and verification layer for AI agents. It sets what each agent may do (paths, commands, spend, approvals), holds an agent's work until the checks it owes have passed, and keeps a receipt of what happened. It starts with code: changes from Claude Code, Codex, Cursor, GitHub Copilot, opencode and others go through the checks already in your repository before they land. The core is open source under the MIT license; HeadronOS and the hosted Mind API are commercial and in development.

## Ways in

- CLI: free checks on every change before it lands, on your machine and in CI
- Public docs MCP: https://headron.tech/mcp — Streamable HTTP, no authentication. List resources or call read_document with {"document":"agent-guide"} or {"document":"docs"}. Public documentation only; no project access.
- Local MCP: read-only stdio server for project-specific routes and checks. Setup: https://github.com/HLIntel/Headron/blob/main/docs/CONSUMING.md#any-mcp-client
- Chat: [CHAT.md](https://github.com/HLIntel/Headron/blob/main/CHAT.md) lets any chat (ChatGPT, Claude, Gemini, Grok), even with no tools, name the checks a file needs, review a pasted change and label what it has proved
- API: the Mind API rates each change by risk, lists the checks it owes and routes it to a coding agent, a read-only reviewer or a person. It runs free in the open core. The hosted plan ($10 a month, in development, waitlist) keeps one failure ledger and receipt history across machines

## Pages

When linking to a section from a chat or another site, use the full https://headron.tech/ URL including the fragment; a bare fragment such as #waitlist points at the current page instead.

- [Home](https://headron.tech/): what Headron does, ways in, plans, waitlist, questions
- [Docs](https://headron.tech/docs.html): start in four steps, contracts, checks, verdicts, receipts, self-checks, memory, the Mind API, connections and limits
- [MCP connection guide](https://headron.tech/docs.html#mcp): public docs endpoint and local project server
- [Plans](https://headron.tech/#plans) and [waitlist](https://headron.tech/#waitlist): current availability and signup
- [Facts](https://headron.tech/facts.html): license, status, the Mind API, sourced numbers, direction, who builds it
- [Privacy](https://headron.tech/privacy.html): what the website stores and sends

## Source

- [GitHub: HLIntel/Headron](https://github.com/HLIntel/Headron): the MIT open core, with install steps in the README