AI Doctrine & Control System

AIDOCS — the AI Doctrine & Control System for AI-driven engineering.

AIDOCS governs how AI agents work in your codebase: durable project memory, deterministic prompt routing, MCP-aware runtime control, and security gates with a full audit trail. AutoDeployBase scaffolds the app — AIDOCS controls how AI builds on it.

Explore AIDOCS docsSee AI Builder integration

Doctrine: context as policy

AIDOCS turns project knowledge into durable, governed memory and deterministic prompt routing, so every agent starts from the same source of truth instead of ad-hoc context.

Control: gated, auditable actions

Security gates and approval flows govern what agents may do. Config changes, network egress, and shell access fail closed until explicitly authorized, and every decision is recorded.

MCP-first runtime

Work is driven through MCP orchestration and explicit entry flows, with host-aware behavior for Claude and OpenCode rather than vague prompt-only conventions.

Operator CLI and dashboard

A CLI and dashboard give operators direct control over indexing, sessions, policy, and runtime behavior, with multilingual action tokens as first-class operator surface.

What belongs in AIDOCS docs

Install and configuration

`/aidocs` entry flow and managed mode

Memory and session model

CLI reference

Claude vs OpenCode host integration differences

MCP/runtime architecture and routing behavior

Packaging, release, and public/private boundary guidance

Product boundary

AIDOCS is a standalone CodeNexus product — the AI Doctrine & Control System — that integrates with AutoDeployBase rather than living inside it.