Build with ACGP¶

Ready to implement ACGP in your agent systems? This section provides everything you need to build standard-profile alpha governed agents.

Choose Your Implementation Path¶

The Build section is centered on the active Standard alpha implementation path.

Regulated Controls is additive; full claim semantics live in ACGP-6 Conformance.

Standard Profile

Perfect for: Production apps, customer-facing systems

Time to implement: 1-2 days

What you get: - ARS risk scoring and CTQ evaluation - Five primary intervention levels plus orthogonal flag - Basic tripwire support - Framework integrations

Get Started

Regulated Controls Badge

The Regulated Controls Badge is additive to the active Standard claim. See ACGP-6 Conformance.

Compare conformance profiles

Recommended Alpha Execution Path¶

Python is the canonical alpha runtime evaluator.
TypeScript currently provides protocol models, validation surfaces, and client integration value.
Non-Python applications should call a supported evaluator service.

Use this model when you need cross-language deployment: your application emits traces and evidence, the evaluator service forwards them to the Python runtime, and the runtime returns the intervention decision plus audit-ready evidence.

Implementation Guides¶

Step-by-step guides for implementing ACGP:

Quick Start¶

Blueprint Usage - Author a blueprint and evaluate traces through GovernanceSteward
Integration Walkthrough - Full path from install to production verification
Reference Deployment - Docker Compose evaluator service with PostgreSQL-backed audit records
Bootstrap Trust - Minimal trust-root and steward-identity profile for serious alpha deployments
Error Handling - Implement robust intervention and exception handling
Observability - Monitor decisions, trust debt, and audit exports

Conformance¶

Implementing Conformance - Standard implementation and badge-ready operational guidance

Advanced Topics¶

Integration Walkthrough - Build from session bootstrap through production hardening
Reference Deployment - Run the supported alpha deployment path end to end
Bootstrap Trust - Define steward trust roots, identity binding, and fast-path verification
Error Handling - Handle interventions, retries, and fallback defaults
Observability - Monitor decisions, trust debt, and audit flows
Runtime Governance Contracts - Implement Runtime Governance Contracts
Security Hardening - Secure your governance implementation
Performance Tuning - Optimize for your latency requirements
Troubleshooting - Common issues and solutions

Framework Integrations¶

ACGP integrates with popular agent frameworks:

Supported Frameworks¶

LangChain

Add governance to LangChain agents with native integration.

LangChain Guide
LangGraph

Govern multi-agent LangGraph workflows.

LangGraph Guide

Community Integrations¶

Protocol Integrations¶

MCP Integration - Model Context Protocol support
A2A Integration - Agent-to-Agent communication
Custom Protocols - Build your own integrations

View all integrations

Development Workflow¶

Typical ACGP implementation workflow:

graph LR
    A[Assess Agent Risk] --> B[Assign Governance Tier]
    B --> C[Select Conformance Level]
    C --> D[Implement Governance]
    D --> E[Configure Blueprints]
    E --> F[Test & Validate]
    F --> G[Deploy]
    G --> H[Monitor & Tune]
    H -.feedback loop.-> E

Assess - Evaluate your agent's risk profile
Choose - Select appropriate Governance Tier and conformance level
Implement - Follow implementation guide for your tier
Configure - Set up governance policies and blueprints
Test - Validate governance behavior
Deploy - Roll out to production
Monitor - Track interventions and tune policies

Quick Links¶

Code Examples

See working implementations for each conformance level.

View Implementation Guide
API Reference

Complete API documentation and schemas.

API Docs
Get Help

Join our community for support and discussions.

Community

Next Steps¶

Choose your conformance level based on your requirements
Follow the implementation guide for your chosen tier
Integrate with your framework using our integration guides
Test thoroughly before deploying to production
Monitor and tune based on real-world behavior

Need help deciding? Check out ACGP-6 Conformance or ask in Discord.