Objective
One workflow with business outcome, scope, policy, success criteria, and stop conditions.
Platform
The smallest useful AI workforce platform loop.
AgentFoundry is not a greenfield framework. It is a productized, Vercel-ready layer over existing agent runtimes, tools, skills, evals, browser/computer-use, proof packets, and runtime cells.
Runtime
Agents, tools, MCP connectors, browser/code execution, memory, skills, and human checkpoints.
Governance
Tool permissions, approval rules, audit trails, rollback plans, risk flags, and policy gates.
Proof
Traces, evals, artifacts, receipts, handoffs, and a reusable run packet.
Learning
Failures become evals, policies, prompts, skills, and reusable workflow recipes.
VVV: Valuable, Verifiable, Viable.
The platform stays honest by requiring real work, real verification, and a viable deployment path before claiming progress.
Valuable — the agent completes meaningful business work, not just a demo task.
Verifiable — every run emits traces, evals, receipts, approvals, artifacts, and a proof packet.
Viable — built from existing OpenClaw, Hermes-agent, OpenCode, MCP, AG-UI, GitClaw, GitPilot, and repo surfaces.