The CorpAI Codex

The founding philosophy of the CorpAI standard.

Why This Exists

AI agents are multiplying. Developers are wiring them together in ad-hoc ways — no structure, no hierarchy, no protocol. Tasks get dropped. Escalations never reach the right agent. The human owner has no idea what’s happening until something breaks.

CorpAI fixes that.

We define a universal standard for how AI agents organize themselves — who reports to who, how tasks flow, when humans get notified, and what every agent is responsible for. Think of it as the employee handbook for your AI workforce.

Core Principles

1. Every agent has a rank

No agent operates without a level. Rank determines authority, responsibility, and escalation path. See spec/ranks.md.

2. Every agent has a chain of command

Every agent knows exactly who they report to and who reports to them. No ambiguity.

3. Communication is structured

Agents don’t send freeform messages. They send Tasks, Reports, Escalations, and Notifications. Each has a defined format. See spec/communication.md.

4. Escalation is explicit

Every agent knows under what conditions to escalate — and to whom. Nothing silently fails. See spec/escalation.md.

5. The OWNER is always in control

Above all agents sits the human OWNER. They receive escalations that no agent can resolve. They set the mission. They can override anything. See roles/executive/OWNER.md.

6. The spec is open

CorpAI belongs to the community. Anyone can propose new roles, departments, or spec updates via pull request.

What CorpAI Is Not

Version

This is CorpAI v0.1 — the founding version. See ROADMAP.md for what’s next.

“Structure is what separates a workforce from a mob.”