Documentation

Guides and references for the vr.dev SDK, API, and CLI.

Quickstart

Get up and running with the vrdev SDK in under 5 minutes.

Concepts

Understand the 3-tier taxonomy, composition engine, and trust model.

API Reference

Complete REST API documentation for the vr.dev verification service.

Verifier Authoring

Create, test, and publish your own verifiers to the registry.

CLI Reference

Command-line interface documentation for the vr tool.

Integration Guide

Connect vr.dev to TRL, VERL, OpenClaw, and your training pipeline.

Examples

End-to-end demo scripts and benchmark results showing verifiers in action.

Bring Your Own State (BYOS)

Use the pre_result pattern to verify agent work against any state source.

Trust Model

How vr.dev ensures verification integrity through evidence chaining, deterministic checks, and transparency.

Execution Sandbox

Security model for verifier execution: permissions, isolation, and safe defaults.

Cost & Latency Profile

Performance characteristics and cost model for each verification tier.

Local Execution

Run verifiers locally without any HTTP calls - ideal for RL training loops.

Evidence Replay

Re-verify historical episodes from stored evidence records.

End-to-End Walkthrough

A complete worked example: install, verify, compose, and export rewards.

AGENTIC Verifier Maintenance

How AGENTIC verifiers work and how to maintain them as target systems change.

Local vs Hosted

Compare local SDK execution against the hosted API - features, trade-offs, and when to use each.

MCP Server

Use vr.dev verifiers as tools in Claude Desktop, Cursor, and other MCP-compatible clients.

Golden Pipeline Templates

Copy-paste verification pipelines for common agent tasks.