agentforge-production

Automate production runtime management for AI agents with decoupled brain, hands, and session logs.

Updated Apr 8, 2026
One-click install
npx skills add https://github.com/Kingxiao/agentforge --skill agentforge-production
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: agentforge-production
Source: https://github.com/Kingxiao/agentforge/tree/main/agentforge-production
Command: npx skills add https://github.com/Kingxiao/agentforge --skill agentforge-production

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Production runtime for agents, decoupling brain, hands, and sessions to improve reliability, scalability, and security in long-running deployments.

Core Features & Use Cases

  • Brain/Hands/Session decoupling: Separate inference, tool execution, and durable event logs to enable crash recovery and horizontal scaling.
  • Lazy provisioning: Provision sandboxes only when needed to reduce latency and resource usage.
  • Observability: Centralized session events and metrics for reliability, cost, and performance monitoring.
  • Credential isolation: Vault-backed secret management to keep credentials out of execution environments.
  • Scaling patterns: Guidelines for single-process to full decoupled architectures across workloads.

Quick Start

Activate the production runtime by enabling Phase 9 patterns (brain/hands/session decoupling, lazy sandbox provisioning, and observability) for your agent.

Frequently Asked Questions about agentforge-production

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I make my AI agent runtime scalable for production?

To make an AI agent runtime scalable, decouple inference, tool execution, and session logs. This separation enables horizontal scaling and crash recovery for long-running deployments.

What is brain hands and session decoupling in agent production runtimes?

Brain, hands, and session decoupling separates inference, tool execution, and durable event logs. This architecture ensures fault tolerance and enables independent scaling across agent sessions.

How do I isolate credentials when running agents in a sandbox?

To isolate credentials in a sandbox, use vault-backed secret management. This keeps credentials completely out of the execution environment, ensuring secure agent production runtime operations.

When do I need lazy sandbox provisioning for AI agents?

You need lazy sandbox provisioning when reducing latency and resource usage for agent runtimes. It provisions execution environments only when needed, optimizing production scaling and performance.

How do I get observability across multiple agent sessions and tool calls?

Achieve observability across agent sessions and tool calls by using centralized session events and metrics. This monitors reliability, cost, and performance for fault-tolerant production runtimes.

Can I recover an agent session after a crash using event logs?

Yes, you can recover an agent session after a crash using durable event logs. Event-driven session replay restores the exact state through the decoupled session architecture.