# VirusTotal AI (VTAI)

Trust Every Decision. Human or Agent.

The trust layer for autonomous security, embedded in every execution, every agent, every decision.

Minimized API and machine-readable discovery surfaces for agents, IDEs, and autonomous security builders.

## Choose your surface
- **OpenClaw plugin** (VT Sentinel plugin): Auto-registers with VTAI, scans new artifacts, and blocks or quarantines malicious content inside OpenClaw. Install or connect with `openclaw plugins install clawhub:openclaw-plugin-vt-sentinel`.
- **Google Antigravity extension** (VT Sentinel extension): Real-time download protection for Google Antigravity and Open VSX-compatible IDE workflows. Install or connect with `Extension ID: virustotal.vt-sentinel`.
- **Hermes plugin** (Hermes plugin): Reputation checks, advisor context, and artifact provenance for Hermes execution environments. Install or connect with `hermes plugins install king-tero/hermes-virustotal`.

## Core protocol
- **Registration**: `POST /api/v3/agents/register` — Self-serve VTAI identity and vtai_ token.
- **Auth**: `x-apikey: vtai_...` — Required for protected analysis endpoints.
- **Analysis**: `GET /api/v3/files/{id}` — Minimized file intelligence for agent loops.
- **Upload**: `POST /api/v3/files/` — Submit files when the workflow allows it.

## Agent bootstrap
1. Read /llms.txt for the shortest high-signal overview of VTAI and the current integration surfaces.
2. Register the agent at POST /api/v3/agents/register to obtain a vtai_ token without requiring an existing VirusTotal API key.
3. Use x-apikey: vtai_... for protected VTAI analysis endpoints such as GET /api/v3/files/{id}.
4. Use /skills/BASIC.md, /openapi.json, or the AI plugin manifest to wire VTAI into the agent runtime you control.

## Machine-readable entry points
- **llms.txt** (Best first read): `/llms.txt` — Fastest human-readable and model-readable overview of VTAI, the API, and current integrations.
- **OpenAPI** (Schema): `/openapi.json` — Machine-readable schema for direct tool integration, code generation, and client validation.
- **API Catalog** (API map): `/.well-known/api-catalog` — RFC 9727 catalog exposing the VTAI API surface, OpenAPI description, docs, and health endpoint.
- **Agent Skill** (Agent spec): `/skills/BASIC.md` — Canonical skill sheet for agents that need the minimal protocol and endpoint guidance.
- **Agent Skills Index** (Skill index): `/.well-known/agent-skills/index.json` — Discovery index for published VTAI agent skills using the .well-known Agent Skills draft format.
- **AI Plugin Manifest** (Manifest): `/.well-known/ai-plugin.json` — Manifest describing the API surface and no-friction registration flow for agents.