# CASAN Enterprise / Governed Agent Console (Level 4) · **FUTURE / structure-only**

> Status: **FUTURE.** Documented + scaffolded, NOT implemented in this task.
> `package-release.sh enterprise` **refuses to build** (no fake-complete package) — several
> building blocks already exist inside the core harness, but the governed console itself
> does not. Do not ship this as a product yet.

Enterprise governed AI-SDLC console. Packages: `casan-enterprise`, `casan-governed-console`.

## Building blocks that ALREADY exist (in core harness, reusable)
| Capability | Where |
|---|---|
| RBAC | `packages/casan-harness/scripts/bash/rbac-check.py` |
| Tenant isolation | `tenant-store.sh`, `tenant-paths.sh`, `tenant-registry-verify.sh`, `tenant-crypt.sh` |
| Approval workflow | `approval-verify.sh`, `approval-sign.sh`, `approval-jwt-mint.py` |
| KMS | `vault-kms.sh` |
| WORM ledger | `worm-ledger.py` |
| Kill-switch / quota | `kill-switch.sh`, `cost-spike-detect.sh` |
| Loop governance | Plan-17 loop primitives (`loop-*.py`) |

## Components still to build (NOT in this task)
- Governed Chat Console (Plan-18) · Prompt Mode Router · Model Provider Management
- Operator mode · Codegen mode · Agent/Skill Registry · policy-versioning UI

## Why it refuses to package
Per the packaging principle, a level that isn't implemented must **fail clearly** rather
than emit a fake-complete artifact. Enterprise is `status: future` in
`packaging/levels.json`, so `scripts/package-release.sh enterprise` exits non-zero with an
explanation. When the console is built, flip its status to `preview`/`implemented`.

## To implement later
Sequence: Plan-14 (RBAC console) → Plan-13 (Control Plane) → Plan-18 (Governed Chat
Console: read-only → operator → chat-as-loop → multi-tenant). Reuse the existing blocks
above instead of re-writing them.