@@ -84,7 +84,7 @@ Recheck these observations before acting:
 - Existing Commons run 49092854-f9bd-4b7a-8d12-4ac5ea0e5f70 retains the approved,
   unmerged safe-timer change from job 811bdb98-bcbb-4b65-80b9-f19693aef6b4.
   Preserve it. 29 earlier lab experiments were verified before this release work.
-- Public source: antropos17/commit-zero at 9a3fa76 (05:26 UTC), CI green; use clean .cache/publication and preserve histories.
+- Public source: antropos17/commit-zero at f6cc442 (05:44 UTC), CI green; use clean .cache/publication and preserve histories.
   README GIF is a 15.8s condensed walkthrough of retained Sequence Kit work,
   not a continuous live recording. Static fallback included; desktop/mobile and all
   three CI jobs passed. Evidence: runtime/verification/github-readme-gif-2026-10-19.json.
 
@@ -8,17 +8,17 @@ tracks dated decisions and receipts; this file is the operational entry point.
 ## Repositories and state
 
 - Working repository: `X:\ai-github`, branch `codex/development-foundation`,
-  latest fully checked local commit `0e028aa` at this checkpoint. The local data and model
+  latest fully checked local commit `791e4b6` at this checkpoint. The local data and model
   installation are here. Recheck status before acting; this file can age.
 - Public repository: [antropos17/commit-zero](https://github.com/antropos17/commit-zero),
-  `main` at `9a3fa763d8e67b487a50da6d416a922c562417be` after a normal push.
+  `main` at `f6cc44257d9de68444ad4bd359a4ee26865f6bc4` after a normal push.
   Its separate checkout is `.cache/publication`. The histories are unrelated;
-  the public tree matches local `0e028aa` (`25202319fa6e12539b79cd41515d6f0502656ae4`).
-  Mobile catalog focus, project recovery, and bookmark rate-limit feedback are pending a full gate.
+  the public tree matches local `791e4b6` (`d7e0c1c04be4abc8d1a12252f3ae1dad0f034efa`).
+  The read-only Commons example and archive/export browser coverage are pending a full gate.
   Do not merge histories, reset the local branch or force push. Recheck remote
   before publishing anything else.
-- Public CI [run 39204566088](https://github.com/antropos17/commit-zero/actions/runs/39204566088)
-  passed all three jobs for `9a3fa76`: full gate, Chromium and runtime
+- Public CI [run 39205979750](https://github.com/antropos17/commit-zero/actions/runs/39205979750)
+  passed all three jobs for `f6cc442`: full gate, Chromium and runtime
   dependency audit.
 - This mission has not run local or hosted model inference, spawned product
   agents, submitted promotional posts or created a new release. Prior retained
@@ -86,19 +86,23 @@ Timer proposal from job `811bdb98-bcbb-4b65-80b9-f19693aef6b4`. Preserve it.
    Node tests from listed Playwright specs. A pending
    [hosted-provider draft](PROVIDER-DRAFTS.md) prepares disconnected text-only
    request/response contracts for Anthropic, OpenAI, Google and Mistral.
-7. The pending mobile catalog change focuses and scrolls to selected role
+7. The mobile catalog focuses and scrolls to selected role
    requirements at 390px. A Chrome duplicate-project fixture confirms form
    preservation and address correction without inference. A bookmark HTTP 429
    now has a persistent local error and a retry that can reach HTTP 200.
    Browser specs use two serial fresh-service groups so the ordinary 180/minute
    limiter is not exhausted by earlier test files.
+8. A pending Environment section presents the source-controlled seed-17
+   [Commons fixture](SIMULATIONS.md) as read-only, explicitly simulated evidence.
+   The simulation test compares this JSON to a fresh HTTP reproduction. A
+   Chrome model-free archive/export/restore journey retains its first issue.
 
 ## Evidence and limits
 
-- Latest local frozen full gate at `0e028aa`: 66/66 Node tests, compiled DOM/axe
+- Latest local frozen full gate at `791e4b6`: 66/66 Node tests, compiled DOM/axe
   2/2, build and static serving passed. Stable source hash
-  `571d5118304357718a507c3fe0b7384f645288d4b4b6e45efde76c923b8bc78b`;
-  receipt `runtime/verification/provider-drafts-full-2026-10-19.json` (ignored by Git).
+  `14727701ef13a699e8826003c2658ea4255ff24fab3cafa810b18d5a415be5b4`;
+  receipt `runtime/verification/mobile-recovery-full-2026-10-19.json` (ignored by Git).
 - Installed Chrome browser suite passed 7/7 at desktop and 390px. Real captures
   are `runtime/verification/commons-first-run-{desktop,narrow}-2026-10-19.png`,
   `commons-provider-offline-desktop-2026-10-19.png` and catalog captures. Tests
@@ -112,12 +116,14 @@ Timer proposal from job `811bdb98-bcbb-4b65-80b9-f19693aef6b4`. Preserve it.
   were published at `0ae079d`, with all three public CI jobs passing. The
   disconnected hosted-provider drafts passed the full gate and public CI at
   `9a3fa76`; they have no credential, transport or runtime qualification. The
-  pending mobile catalog and duplicate-address browser unit passed 2/2 targeted
-  Chrome tests, and `check:ui` passed. One shared-cluster Chrome run failed 7/8
-  on a genuine HTTP 429; fresh service groups passed 8/8 twice with the normal
-  limiter. The bookmark 429 → 200 fixture passed at 390px with axe. Its actual
-  [browser capture](media/bookmark-rate-limit-2026-10-19.png) is a routed
-  disposable-server failure, not a model or load result. The full gate remains.
+  mobile catalog/bookmark recovery passed grouped Chrome 8/8 and the public
+  full gate at `f6cc442`. A prior shared-cluster run failed 7/8 on genuine
+  HTTP 429; fresh service groups retain the normal limiter. The bookmark
+  429 → 200 fixture passed at 390px with axe. Its actual [browser capture](media/bookmark-rate-limit-2026-10-19.png)
+  is a routed disposable-server failure, not a model or load result. A pending
+  [Environment capture](media/environment-fixture-2026-10-19.png) shows the
+  read-only fixture. `check:commons` passed 24/24, `check:ui` passed, and grouped
+  Chrome passed 9/9 including archive/export/restore. The full gate remains.
 - `npm run benchmark:backend` used a disposable PGlite snapshot of five projects
   and 1,000 messages. Twenty samples per path: current normalized message-write
   p50 9.09 ms versus retained legacy pattern 42.99 ms. This verifies a prior
@@ -131,7 +137,7 @@ Timer proposal from job `811bdb98-bcbb-4b65-80b9-f19693aef6b4`. Preserve it.
 
 ## Next work, in order
 
-1. Complete the mobile catalog/browser unit: run grouped Chrome and `npm run check` on frozen source,
+1. Complete the fixture UI/archive unit: run focused Commons/UI, grouped Chrome and `npm run check` on frozen source,
    commit it, copy the unpublished local commit into the separate public
    checkout, compare Git trees, push without force and inspect public CI.
    Record exact SHA and receipt.
@@ -163,7 +169,7 @@ Timer proposal from job `811bdb98-bcbb-4b65-80b9-f19693aef6b4`. Preserve it.
 > `docs/HANDOFF-2026-10-19.md`, and `docs/MISSION-2026-10-19.md`. Recheck Git
 > state and `npm run doctor`; use `npm run map -- <area>`. Preserve all runtime
 > data, credentials, earlier model receipts, and separate Git histories.
-> First confirm public CI for `9a3fa76` and complete any dirty mobile catalog or
+> First confirm public CI for `f6cc442` and complete any dirty fixture UI or
 > handoff unit. Then reproduce and improve one
 > concrete browser or transaction journey, using focused checks and the full
 > `npm run check` on frozen source. Keep model-backed catalog roles disabled;
 
@@ -12,7 +12,7 @@ Start: 03:46:17 UTC. Deadline: 15:46:17 UTC. Final integration reserve begins 14
 
 ## Current action
 
-Finish the mobile catalog, duplicate-project and bookmark rate-limit recovery unit. Run the grouped full Chrome suite and frozen full gate; commit and publish after tree comparison. Continue inspecting real user journeys.
+Finish a read-only seeded Commons evidence section and browser coverage for project archive, export and restore. Run focused Commons/UI and grouped Chrome, then the frozen full gate. Commit and publish with tree comparison.
 
 ## Decisions and limits
 
@@ -50,7 +50,9 @@ Finish the mobile catalog, duplicate-project and bookmark rate-limit recovery un
 - 05:30 UTC: inspection of the actual 390px catalog layout found that details are placed above the list; selecting a role from lower in the list left the newly selected details offscreen. The selected heading now receives focus and instant scroll on narrow widths; desktop stays side by side. A new Chrome scenario verified duplicate project-address error preserves name/description and succeeds after correcting the address. Targeted Chrome checks passed 2/2 with axe and no page errors. Focused UI, full browser and full gate remain.
 - 05:35 UTC: `check:ui` passed, then the expanded one-cluster Chrome suite failed 7/8: bookmark PUT returned real HTTP 429 after preceding fixture requests. Isolated bookmark spec passed. A second instrumented full run confirmed 429. Browser specs now run in two serial fresh-service groups, each retaining the normal 180/minute limiter; full Chrome passed 8/8. The bookmark control now retains a visible local error for 429 and supports an explicit retry; a routed 429 → real 200 fixture passed alone.
 - 05:42 UTC: after adding narrow/axe coverage to the HTTP 429 state, `check:ui` and the grouped Chrome suite passed again (8/8). Actual 390px capture `docs/media/bookmark-rate-limit-2026-10-19.png` matches ignored runtime evidence by SHA-256 `b842ad70`; its fixture route returns 429 only once, then the local service returns 200. `npm run diagnostics`: 83 GB free, `.cache` 918 MB, retained stores/backups/models preserved. Full frozen gate remains.
+- 05:44 UTC: mobile/error-recovery unit committed locally as `791e4b6` after frozen full gate 66/66 (source hash `14727701`). All 19 changed files matched by SHA-256 and the Git trees matched at `d7e0c1c`; normal push advanced public `main` to `f6cc442`. [CI run 39205979750](https://github.com/antropos17/commit-zero/actions/runs/39205979750) passed full gate, Chromium and runtime dependency audit.
+- 05:47 UTC: real Chrome verified a model-free project can be archived, exported as JSON with its issue intact, then restored and reloaded. The shared seed-17 Commons example is now a read-only Environment section with explicit fixture/no-inference labels. Its committed JSON matches a fresh HTTP simulation in the existing test; targeted Chrome/axe checked the 390px view. Actual capture `docs/media/environment-fixture-2026-10-19.png` matches ignored runtime PNG by SHA-256 `9d88aafe`. `check:commons` passed 24/24, `check:ui` passed, and grouped Chrome passed 9/9. Full gate remains.
 
 ## Next checkpoint
 
-Run frozen `npm run check` for this UI/recovery unit, then commit and publish through the separate checkout after tree comparison. Continue bounded implementation. Keep the final 90-minute reserve from 14:16:17 UTC.
+Run frozen `npm run check` for simulation evidence and archive/export coverage, then commit and publish through the separate checkout after tree comparison. Continue bounded implementation. Keep the final 90-minute reserve from 14:16:17 UTC.
@@ -29,7 +29,15 @@ carry `test_fixture` provenance. The script asserts, rather than merely prints:
 `node --test tests/simulate-commons.test.mjs` replays the same seed twice and
 compares the normalized reports. It runs a larger history with another seed and
 checks a distinct merged source hash and additional cursor pages. Invalid seeds
-and history sizes are rejected before opening a store. The fixture code is a source-controlled reproduction tool, not evidence
+and history sizes are rejected before opening a store. The seed-17 output is
+stored as `shared/commons-fixture-seed17.json`, and the same test compares it
+with a fresh HTTP simulation result. The Environment page displays its selected
+outcomes under **Seeded Commons example** with fixture actor and no-inference
+labels. It is a read-only sample, separate from both the retained user store
+and the original scenario. This [actual narrow Chrome capture](media/environment-fixture-2026-10-19.png)
+shows that view; the sample was generated by the command above, not by a model.
+
+The fixture code is a source-controlled reproduction tool, not evidence
 of model quality or autonomous collaboration. It neither executes generated
 JavaScript nor qualifies an arbitrary-code sandbox. For actual model history,
 read the preserved Commons run and its inference receipts separately.
 
@@ -3221,6 +3221,15 @@ textarea::placeholder {
 .sandbox-layout .service-list {
   margin-bottom: 20px;
 }
+.commons-fixture-summary {
+  margin-top: 24px;
+  padding-top: 20px;
+  border-top: 1px solid var(--border);
+}
+.commons-fixture-summary h2 {
+  font-size: var(--type-section);
+  margin: 0 0 12px;
+}
 .search-section:first-child {
   margin-top: 0;
 }
 
@@ -13,6 +13,7 @@ import React, { useState } from "react";
 import { useAccount, useResource, useWorkspace } from "../hooks.mjs";
 import { path, replaceRoute } from "../router.mjs";
 import { stages, before, after } from "../../../shared/scenario.mjs";
+import commonsFixture from "../../../shared/commons-fixture-seed17.json";
 import { TimelineFeed } from "./timeline.jsx";
 import {
   Avatar,
@@ -1183,6 +1184,23 @@ export function Environment() {
             Docker daemon. Runtime denial, timeout, cancellation, and artifact
             capture must pass before activation.
           </p>
+          <section className="commons-fixture-summary" aria-label="Seeded Commons example">
+            <h2>Seeded Commons example</h2>
+            <p>
+              Fixture only · seed {commonsFixture.seed} · no model inference or
+              saved user-history write. These outcomes come from the isolated
+              Commons HTTP reproduction.
+            </p>
+            <p>{commonsFixture.actorLabels.join(" · ")}</p>
+            <ul className="plain-list">
+              <li>Stale edit: HTTP {commonsFixture.outcomes.staleRevisionStatus}; no extra write.</li>
+              <li>Rejected review: {commonsFixture.outcomes.rejectedReview}; merge HTTP {commonsFixture.outcomes.rejectedMergeStatus}.</li>
+              <li>Approved fixture merge: project v{commonsFixture.outcomes.acceptedMergeVersion}.</li>
+              <li>Failure, explicit retry and cancellation: {commonsFixture.outcomes.providerFailure}, {commonsFixture.outcomes.explicitRetry}, {commonsFixture.outcomes.cancelled}.</li>
+              <li>{commonsFixture.outcomes.events} events across {commonsFixture.outcomes.pages} cursor pages; interrupted job recovered.</li>
+            </ul>
+            <p>Reproduce: <code>npm run simulate:commons -- --seed {commonsFixture.seed}</code></p>
+          </section>
         </section>
         <aside className="page-support" aria-label="Execution boundaries">
           <h2>Boundaries</h2>
 
@@ -39,6 +39,7 @@ export const areas = {
       "shared/agent-catalog.mjs",
       "shared/agent-evaluation.mjs",
       "shared/provider-drafts.mjs",
+      "shared/commons-fixture-seed17.json",
       "db/migrations/003_commons.sql",
       "db/migrations/004_commons_storage.sql",
       "db/migrations/005_commons_issues.sql",
 
@@ -0,0 +1,30 @@
+{
+  "kind": "deterministic-commons-fixture",
+  "seed": 17,
+  "historyCount": 8,
+  "modelInference": false,
+  "persistedUserData": false,
+  "actorLabels": [
+    "simulated operator",
+    "Fixture Ada (no model)",
+    "Fixture Ben (no model)",
+    "simulated reviewer"
+  ],
+  "outcomes": {
+    "staleRevisionStatus": 409,
+    "unavailableCapabilityStatus": 503,
+    "rejectedReview": "changes-requested",
+    "rejectedMergeStatus": 409,
+    "acceptedMergeVersion": 1,
+    "acceptedSourceHash": "8a80aa465ff0ac7b017e240efc7fcbb93d744b1f8a139839bfa9a418957d318e",
+    "providerFailure": "failed",
+    "explicitRetry": "completed",
+    "invalidOutput": "failed-without-pr",
+    "timeout": "failed-without-pr",
+    "cancelled": "cancelled",
+    "events": 48,
+    "pages": 16,
+    "scenarioRevisionPreserved": true,
+    "interruptedRecovery": true
+  }
+}
@@ -50,6 +50,12 @@ test("disabled role catalog stays inspectable at desktop and narrow widths", asy
   await page.goto("/#/sandbox");
   await expect(page.getByRole("heading", { name: "Code runner readiness" })).toBeVisible();
   await expect(page.locator("main")).toContainText("Docker execution is not connected");
+  await expect(page.getByRole("region", { name: "Seeded Commons example" }))
+    .toContainText("Fixture only · seed 17 · no model inference");
+  await expect(page.getByRole("region", { name: "Seeded Commons example" }))
+    .toContainText("48 events across 16 cursor pages");
   await expect(page.locator("main button")).toHaveCount(1);
+  if (process.env.CZ_CAPTURE_BROWSER === "1")
+    await page.screenshot({ path: "runtime/verification/environment-fixture-narrow-2026-10-19.png", fullPage: true });
   expect(errors).toEqual([]);
 });