Keep your agents behaving the way you taught them, on every model and tool.

The files that steer AI agents, CLAUDE.md, AGENTS.md, GEMINI.md, SKILL.md, Cursor rules, Copilot instructions, Windsurf rules, fail silently. They drift, and nothing tells you until the agent does the wrong thing. TomeVault watches how your setup behaves across models and reconciles it as they change. We are owned by no model vendor and we sell none of the tools we grade.

tomevault.io

Scan. Drop in a file and read the verdict. Does it load, is it clear enough to follow, is it safe, and does it hold up across models, with the exact fix for anything that's off. Every check is a deterministic pattern matcher against a public ruleset, never a model judging a model, so you can read the rule that made the grade rather than trust it. No sign-up, nothing uploaded.

Watch. A scan checks your setup once. Watchdog keeps it true. It watches the repo where your instructions live and re-checks them on every push and every model release, catches what drifted and fixes it in a pull request you merge, and signs the result. It acts when something changes, not on a schedule.

Sign. A verified setup carries a signed record in its repo: signed_by: null becomes signed_by: tomevault. tome.lock catches a changed byte. Watchdog catches a changed model. The signature makes the verdict portable, so anyone can prove the setup was graded against a public standard and is current.

Translate. Behaviour authored once, made to run everywhere. A file written for one agent renders to all of them and stays in sync, across Claude Code, Codex, GitHub Copilot, Cursor, Gemini CLI, and Windsurf. The easy part, handled.

Vault. A Vault gives a team one shared, graded, signed home for the instruction files their AI tools rely on, so every teammate's agents behave the way the team agreed. tomevault.io/watchdog

Symlinks keep your files looking the same. They can't keep your agents behaving the same.

Agent conformance is proving your AI agents are still running the rules you approved. Evals score what an agent did. Conformance proves what it was told is still true. The standard lives at tomevault.io/standards/agent-conformance.

• Run a free Diagnostic on any instruction file: tomevault.io/diagnostic
• Keep it true automatically: tomevault.io/watchdog
• A signed credential for your clients: tomevault.io/agencies
• The standard behind every verdict: tomevault.io/standards
• Read the docs: tomevault.io/docs

One source of truth for your AI instructions. tomevault.io