docs: add submission screenshots

README.md

@@ -27,6 +27,10 @@ one of only two outcomes:
 > ProofGate never deploys, rolls back, changes a flag, or writes to production.
 > It prepares evidence for a human decision.
 
+![ProofGate returning READY_FOR_HUMAN after three concurrent evidence checks](docs/assets/control-room-ready.png)
+
+<p align="center"><em>Three independent evidence lanes, one case digest, and a human-owned verdict.</em></p>
+
 ## Why concurrency is the point
 
 This is not a sequential pipeline presented as a group of agents. Three
@@ -132,6 +136,27 @@ All scenarios are deterministic and synthetic. Running them needs no model API
 key, production credential, registry access at runtime, or private Fortemate
 data.
 
+## Visual proof
+
+| Fail closed on unsafe evidence                                                                                      | Observe all three agents in Mozaik Cloud                                                             |
+| ------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------- |
+| ![ProofGate blocking a release candidate after an experiment guardrail fails](docs/assets/control-room-blocked.png) | ![Three ProofGate evidence agents visible in Mozaik Cloud](docs/assets/mozaik-cloud-concurrency.png) |
+
+<details>
+<summary><strong>Inspect the event ledger and one agent trace</strong></summary>
+
+The ledger preserves the interleaved lifecycle events and distinct loop IDs used
+to verify overlap:
+
+![ProofGate event ledger with interleaved events from three agents](docs/assets/event-ledger.png)
+
+Mozaik Cloud exposes each agent's constrained instruction, tool, and synthetic
+context for debugging:
+
+![Mozaik Cloud details for the policy evidence agent](docs/assets/mozaik-cloud-agent-details.png)
+
+</details>
+
 ## Trust model
 
 The agent participants coordinate work, but they do not decide the gate. The

docs/assets/README.md

@@ -6,7 +6,8 @@ Expected filenames:
 
 - `control-room-ready.png`
 - `control-room-blocked.png`
+- `event-ledger.png`
 - `mozaik-cloud-concurrency.png`
-- `event-ledger.png` (optional)
+- `mozaik-cloud-agent-details.png` (optional)
 
 See [`../demo-script.md`](../demo-script.md) for framing and redaction guidance.

docs/demo-script.md

@@ -61,9 +61,11 @@ Place final images in `docs/assets/` with these exact names:
    cards, the case digest, and `overlap verified` visible.
 2. `control-room-blocked.png` — the `BLOCKED` result with the experiment reason
    visible.
-3. `mozaik-cloud-concurrency.png` — the Cloud Agents view showing three traces
-   from one fresh run, with any account details or keys cropped out.
-4. `event-ledger.png` — optional closer crop of the expanded interleaved ledger.
+3. `event-ledger.png` — a closer crop of the expanded interleaved ledger.
+4. `mozaik-cloud-concurrency.png` — the Cloud Agents view showing all three live
+   agent definitions and their loops, with any account details or keys cropped out.
+5. `mozaik-cloud-agent-details.png` — an optional Cloud detail view showing one
+   agent's restricted instruction, tool, and synthetic context.
 
 Prefer PNG or WebP, use a 16:9 aspect ratio where practical, and keep each image
 below roughly 1.5 MB. Do not include credentials, project keys, email addresses,