# Five-Minute Live Enterprise Demonstration Scenario (FIRME AI CORPORATION)

1. **Minute 0–1 (The Threat):** Show an autonomous agent attempting an unauthorized tool call (`shell_exec`), a replayed permit, and a parameter-tampered `read_dataset` call (`iris` → `/etc/shadow`).
2. **Minute 1–2 (MCPGate Enforcement):** Run `python3 -m cain.e120.governed_mission` live: `MCPTransparentInterceptor` + Ed25519 capability token verification rejects 8/8 unauthorized/tampered/replayed calls fail-closed while allowing the single authorized call.
3. **Minute 2–3 (Live Multi-Process Kill-Switch):** Show 3 active agent workers hammering the resource while a separate OS process triggers `KILL`; inspect the resource's hash-chained ledger showing **0 effects after `KILL_FENCE`**.
4. **Minute 3–4 (Independent Offline Verification):** Download any public bundle from `https://clawx.click/evidence/` and run `python3 -I verify_e120_offline.py .` (zero CAIN imports, pure Python stdlib) to verify every SHA-256 hash and statistical p-value offline.
5. **Minute 4–5 (Next Step — $500 Assessment):** Walk through `https://cainstudio.online/assessment/` — a 3-day fixed-scope security & governance assessment of the buyer's actual MCP tools/policies.
