AgentV

Documentation

Choose Your Path

Industry Evaluator

The Commander
Explore sector-specific Taxonomies, deploy Industrial Benchmarks, and manage Drift.

Agent Integrator

The Navigator
Connect your agent via REST/Socket, integrate with Frameworks, and run your First Eval.

Platform Extender

The Alchemist
Build World Shims, write Custom Metrics, and extend the CLI & Plugins.

Security Auditor

The Guardian
Verify Trust Protocols, audit Compliance, and enforce Hardened Security.

OSS Core Builder

The Architect
Master the Core Engine, extend Framework Adapters, and optimize Platform Performance.

The Scholar

The Researcher
Standardize on NIST AI-100-1, leverage Deterministic Seeds, and execute Industry Benchmarks.

---

🕵️ Forensic Walkthroughs

Gain instant gratification by exploring industrial-grade evaluation scenarios and their forensic results.

Fintech Decathlon

High-Stakes Loan Decisioning
A complete end-to-end walkthrough of a credit-risk agent navigating complex regulatory checks and multi-step tool use.

Healthcare Audit

HIPAA Compliance Verification
Verifying agent behavior against PHI data masking and audit-log integrity protocols.

---

💎 Core Industrial Features

• Zero-Touch Extensibility: Hot-swappable engine and CLI extension discovery via Python Entry Points.
• VFS-Aware Sandboxing: 20+ world shims (Git, DB, Slack, CRM) for high-fidelity environment simulation.
• Behavioral DNA: High-granularity telemetry capturing the internal reasoning paths of complex agents.
• NIST AI-100-1 Alignment: Standardized trustworthiness scoring across 7 critical AI dimensions.
• Asymmetric Trust Protocol: ED25519-based non-repudiable audit trails and cryptographic result verification.
• Interactive Visual Suite: Premium React dashboard with real-time trajectory playback and forensic triage.

---

🏁 Instant Gratification

Get up and running in under 60 seconds with the built-in demo agent:

pip install -e .
agentv quickstart