scoreboard: the landing demo's four OpenEMR referrals, each report under an offline Seal

packages/flow/benchmark/scoreboard/raw/openemr_referral/seals/2026-10-19/README.md

@@ -0,0 +1,27 @@
+# Referral demo runs, with offline Seals
+
+The compiled OpenEMR referral program on the landing page's four referrals,
+run on 19 October 2026 on a private copy of the pinned OpenEMR fixture. All
+patients are synthetic. No model was called.
+
+| Referral | Patient | Run | Outcome in the report | Read-back |
+|---|---|---|---|---|
+| 1 | Nadia Brennan | `referral-1-r1-nadia-brennan/engine` | `COMPLETED_UNVERIFIED`, every record check confirmed | 1 new chart, 12 of 12 fields |
+| 2 | Marcus Feld | `referral-2-r2-marcus-feld/engine` | `COMPLETED_UNVERIFIED`, every record check confirmed | 1 new chart, 12 of 12 fields |
+| 3 | Ines Carvalho | `referral-3-r3-ines-carvalho/engine` | `COMPLETED_UNVERIFIED`, every record check confirmed | 1 new chart, 12 of 12 fields |
+| 4 | Rosa Delgado, 1958-03-22 (possible duplicate of the seeded chart born 1958-03-27) | `referral-4-r6-rosa-delgado/engine` | `HALTED` at the duplicate check (`step_039`: "No matches were found." not shown) before saving; transaction outcome `RECONCILIATION_REQUIRED` | no new chart |
+
+- **What ran.** The #1269 harness's own run (`benchmark/openemr_api_tier/feature_e2e.run_one`, engine 93c32e9f), with API steps off (the screen program). The engine's record check reads OpenEMR's REST API through the fixture's read-only oracle client. The runs use no named execution profile, so a confirmed run is recorded as `COMPLETED_UNVERIFIED`, not `VERIFIED`. `VERIFIED` needs the standard profile, whose gate refuses this bundle today.
+- **Referral 4 is not `HALTED_BEFORE_EFFECT`.** Nothing was saved, but steps whose risk was never reviewed keep the engine from proving that, so it records `RECONCILIATION_REQUIRED` (see #1331).
+- **Seals.** Each `seal.json` is an **offline Seal, signed by this machine's key** (`openadapt.local-seal/v1`, key `seal-ed25519-accd03a9246d29bb`). It covers the exact bytes of the `report.json` beside it and copies that report's outcome; it never upgrades it. It is not an OpenAdapt Cloud Seal. A statement signed by the machine that ran the work is weaker evidence than an independent party's.
+- **Referral 4 has a report and its offline Seal, and no receipt.** The local receipt is written only for a `VERIFIED` run.
+
+Verify any run against the pinned public key in `seal-key.json`:
+
+```bash
+openadapt-flow local-seal verify referral-1-r1-nadia-brennan/engine --keys seal-key.json
+```
+
+It prints `valid offline Seal over report.json, key seal-ed25519-accd03a9246d29bb`. A changed report prints `not valid: subject_mismatch`.
+
+`summary.json` lists each run's outcome, read-back and Seal digest. `result.json` beside each run is the harness's own SQL read-back. Screenshots are not kept. The script is `scoreboard/seals_referral_demo.py`.

packages/flow/benchmark/scoreboard/raw/openemr_referral/seals/2026-10-19/referral-1-r1-nadia-brennan/engine/seal.json

@@ -0,0 +1,13 @@
+{
+  "algorithm": "ed25519",
+  "issued_at": "2026-10-20T01:11:03Z",
+  "issuer": "local",
+  "key_id": "seal-ed25519-accd03a9246d29bb",
+  "outcome": "COMPLETED_UNVERIFIED",
+  "public_key": "JpfLXYLyY0CVqHz4tHY8eTyC9e7ggpoxLPq6ITgDeHU=",
+  "run_id": "engine",
+  "schema": "openadapt.local-seal/v1",
+  "signature": "LMkyYwygJaPBVzaTIrd4ZO2jE3TSu8YO5ODn+wSIUvsCv5/eV4tdS4Bl9FaoWAeYkuSWKcsfsQH76XDH9IVRDw==",
+  "subject_kind": "run-report",
+  "subject_sha256": "be376829e0e3e8276423a8a7f6f151a45f85aab12476226c70faabe73067cf8d"
+}

packages/flow/benchmark/scoreboard/raw/openemr_referral/seals/2026-10-19/referral-1-r1-nadia-brennan/result.json

@@ -0,0 +1,94 @@
+{
+ "arm": "browser",
+ "ref": "r1-nadia-brennan",
+ "mismatch": false,
+ "error": null,
+ "engine_success": true,
+ "outcome": "COMPLETED_UNVERIFIED",
+ "transaction_outcome": "COMPLETED_UNVERIFIED",
+ "actuation": "dom",
+ "api_fallback": "API steps are off for this run",
+ "steps": 41,
+ "step_errors": [],
+ "new_charts": 1,
+ "readback": "12/12",
+ "correct": true,
+ "field_check": [
+  {
+   "field": "title",
+   "want": "Ms.",
+   "got": "Ms.",
+   "match": true
+  },
+  {
+   "field": "fname",
+   "want": "Nadia",
+   "got": "Nadia",
+   "match": true
+  },
+  {
+   "field": "lname",
+   "want": "Brennan",
+   "got": "Brennan",
+   "match": true
+  },
+  {
+   "field": "DOB",
+   "want": "1971-04-19",
+   "got": "1971-04-19",
+   "match": true
+  },
+  {
+   "field": "sex",
+   "want": "Female",
+   "got": "Female",
+   "match": true
+  },
+  {
+   "field": "street",
+   "want": "27 WILLOW BEND ROAD",
+   "got": "27 WILLOW BEND ROAD",
+   "match": true
+  },
+  {
+   "field": "city",
+   "want": "EXAMPLETON",
+   "got": "EXAMPLETON",
+   "match": true
+  },
+  {
+   "field": "state",
+   "want": "MA",
+   "got": "MA",
+   "match": true
+  },
+  {
+   "field": "postal_code",
+   "want": "02139",
+   "got": "02139",
+   "match": true
+  },
+  {
+   "field": "phone_home",
+   "want": "555-0121",
+   "got": "555-0121",
+   "match": true
+  },
+  {
+   "field": "email",
+   "want": "nadia.brennan@example.invalid",
+   "got": "nadia.brennan@example.invalid",
+   "match": true
+  },
+  {
+   "field": "country_code",
+   "want": "USA",
+   "got": "USA",
+   "match": true
+  }
+ ],
+ "wall_s": 24.978,
+ "setup_s": 1.734,
+ "engine_s": 21.733,
+ "readback_s": 1.512
+}