feat(studio): data-repo registry — lab repo picker, "Rig id", no default repo (v0.72)

CLAUDE.md

@@ -214,7 +214,18 @@ fix flows to every page automatically; two hand-written HTML pages never will.
   firmware's ASCII error payload on non-ok status — keep that for new ops.
   New `.cmenu` popups: the document click-away closer ignores clicks inside
   `.cmenu-pop`; one-shot `.cmenu-item`s (not in a `.cmenu-row`) auto-close.
-- **Metadata / controlled-vocab sourcing (THE rule):** when a **course repo is
+- **Data repos (v0.72):** the Studio/Pattern Designer/dashboard target ONE GitHub
+  data repo chosen in File ▾ → GitHub → Repo, from the registry in
+  `js/studio-data-repos.js` (classic dual-export; course + lab entries + `parseRepo`
+  + label helpers). **No default repo** on a fresh browser. **Copy rule:** user-visible
+  wording goes through `Studio.dataRepoLabel()` / `Studio.idLabel()` /
+  `Studio.sharedLabel()` ("Bench id" for the course repo, "Rig id" otherwise);
+  internal identifiers (`courseSettings`, `refreshCourseMeta`, `fetchCourse*`,
+  `openFromCourseRepo`, `fmOpenCourse`, `studio_bench_id`) keep their historical
+  names — tests anchor on them, do not mass-rename. New repo = registry entry +
+  `scripts/seed-data-repo.sh`; never enable branch protection. Setup guide:
+  `docs/development/data-repo-setup.md`.
+- **Metadata / controlled-vocab sourcing (THE rule):** when a **data repo is
   configured AND signed in**, ALL metadata vocabularies load from that repo (its
   root-level YAML) and their ↗ source links repoint there — the connected repo is
   the source of truth. When not (offline / not signed in), fall back to the

course/cshl-2026/docs/github-overview.md

@@ -20,7 +20,8 @@ Everything lives in one repository: **`reiserlab/cshl-2026-course`** (public sin
 August 2026 — anyone can read it; writing from Arena Studio still needs the bench to
 be signed in once per browser).
 
-Each of the 7 bench rigs has a **bench id** (`bench01` … `bench07`). The repo
+Each of the 7 bench rigs has a **bench id** (`bench01` … `bench07`; the field is
+labelled "Bench id" while the course repo is selected in File ▾ → GitHub → Repo). The repo
 is organized so no two benches ever overwrite each other's files:
 
 ```

dashboard/data-browser/app.js

@@ -399,7 +399,7 @@ function renderCatalog() {
     const visible = new Set(visibleDescriptors().map((item) => item.key));
     if (!state.catalog.length) {
         els.runCatalog.innerHTML = '<div class="empty-state">No runlogs indexed</div>';
-        els.catalogStatus.textContent = 'Open files or connect to the course repository.';
+        els.catalogStatus.textContent = 'Open files or connect to a data repository.';
         els.catalogSearchInput.disabled = true;
         els.renderSelectionButton.disabled = true;
         renderFocusOptions();
@@ -1724,6 +1724,18 @@ window.addEventListener('resize', renderScope);
 
 async function initialize() {
     els.githubRepoInput.value = G.currentRepo();
+    // Offer the known data repos (course + lab) as suggestions; free text still works.
+    const dl = document.getElementById('dataRepoList');
+    const registry = window.StudioDataRepos;
+    if (dl && registry && Array.isArray(registry.DATA_REPOS)) {
+        dl.textContent = '';
+        for (const r of registry.DATA_REPOS) {
+            const o = document.createElement('option');
+            o.value = r.full;
+            o.label = r.label;
+            dl.appendChild(o);
+        }
+    }
     loadAnalysisAxes();
     if (window.location.hostname.endsWith('github.io')) {
         els.plotSourceLink.href =

dashboard/data-browser/github-repo.js

@@ -7,7 +7,14 @@
     const REPO_KEY = 'studio_gh_repo';
     const BENCH_KEY = 'studio_bench_id';
     const FOLDERS_KEY_PREFIX = 'dashboard_runlog_folders:';
-    const DEFAULT_REPO = 'reiserlab/cshl-2026-course';
+    // The registry (js/studio-data-repos.js, loaded before this file) lists the
+    // known data repos; the dashboard's default stays the first entry (the
+    // course repo). Falls back to the literal if the registry failed to load.
+    const REGISTRY = global.StudioDataRepos || null;
+    const DEFAULT_REPO =
+        REGISTRY && REGISTRY.DATA_REPOS && REGISTRY.DATA_REPOS[0]
+            ? REGISTRY.DATA_REPOS[0].full
+            : 'reiserlab/cshl-2026-course';
 
     function currentToken() {
         return sessionStorage.getItem(TOKEN_KEY) || localStorage.getItem(TOKEN_KEY) || '';
@@ -98,7 +105,7 @@
 
     async function apiJson(url, options) {
         const token = currentToken();
-        if (!token) throw new Error('Sign in with the course GitHub token first');
+        if (!token) throw new Error('Sign in with a GitHub token first');
         const response = await fetch(url, {
             method: (options && options.method) || 'GET',
             headers: headers(token),
@@ -125,7 +132,7 @@
         const repo = parseRepo(repoValue || currentRepo());
         const pat = prompt(
             `Paste a GitHub personal access token for ${repo.full} (fine-grained for org members; classic for the shared course account).\n` +
-                'The course token should have Contents read/write access.\n\n' +
+                'It needs Contents read access (read/write keeps it compatible with Arena Studio).\n\n' +
                 'It is stored in sessionStorage first. The next prompt can remember it on this browser.'
         );
         if (!pat) return null;
@@ -134,7 +141,7 @@
         sessionStorage.setItem(TOKEN_KEY, token);
         if (
             confirm(
-                'Remember this token on THIS browser?\nYES for a course bench; NO on a shared personal machine.'
+                'Remember this token on THIS browser?\nYES on a dedicated rig/bench computer or your own laptop; NO on a shared machine.'
             )
         ) {
             localStorage.setItem(TOKEN_KEY, token);
@@ -169,7 +176,7 @@
 
     async function fetchRaw(repoValue, path, ref, prefixBytes) {
         const token = currentToken();
-        if (!token) throw new Error('Sign in with the course GitHub token first');
+        if (!token) throw new Error('Sign in with a GitHub token first');
         const repo = parseRepo(repoValue || currentRepo());
         const requestHeaders = headers(token, 'application/vnd.github.raw');
         if (prefixBytes) requestHeaders.Range = `bytes=0-${Math.max(1023, prefixBytes - 1)}`;

dashboard/data-browser/index.html

@@ -10,7 +10,8 @@
   <script defer src="vendor/kinematics.js?v=20260710-1647"></script>
   <script defer src="analysis-core.js?v=20260710-2236"></script>
   <script defer src="plot-specs.js?v=20260710-2236"></script>
-  <script defer src="github-repo.js?v=20260710-1647"></script>
+  <script defer src="../../js/studio-data-repos.js?v=20260904"></script>
+  <script defer src="github-repo.js?v=20260904"></script>
   <script defer src="app.js?v=20260710-1653"></script>
 </head>
 <body>
@@ -20,7 +21,7 @@
       <h1>Course Data Dashboard</h1>
     </div>
     <div class="header-actions">
-      <a id="courseRepoLink" class="quiet-link" href="https://github.com/reiserlab/cshl-2026-course" target="_blank" rel="noreferrer">Course repo</a>
+      <a id="courseRepoLink" class="quiet-link" href="https://github.com/reiserlab/cshl-2026-course" target="_blank" rel="noreferrer">Data repo</a>
       <div class="status-strip" id="statusStrip" aria-live="polite">
         <span class="status-dot" aria-hidden="true"></span>
         <span id="statusText">No data loaded</span>
@@ -49,7 +50,7 @@ <h2>Data</h2>
         </div>
 
         <div class="github-controls">
-          <input id="githubRepoInput" type="text" value="reiserlab/cshl-2026-course" aria-label="GitHub repository">
+          <input id="githubRepoInput" type="text" list="dataRepoList" value="reiserlab/cshl-2026-course" aria-label="GitHub repository"><datalist id="dataRepoList"></datalist>
           <button id="githubSignInButton" type="button">Sign in</button>
           <button id="githubSignOutButton" type="button" disabled>Sign out</button>
           <button id="chooseRigsButton" type="button" title="Choose which rig folders are included in the data catalog" disabled>Select rigs</button>
@@ -107,7 +108,7 @@ <h2>Data</h2>
       </div>
 
       <div class="catalog-head">
-        <span id="catalogStatus">Open files or connect to the course repository.</span>
+        <span id="catalogStatus">Open files or connect to a data repository.</span>
         <input id="catalogSearchInput" type="search" placeholder="Filter by rig, experimenter, run, fly, genotype, protocol, notes" disabled>
       </div>
       <div id="runCatalog" class="run-catalog" aria-label="Available runlogs">

docs/development/arena-studio-release-notes.md

@@ -4,6 +4,24 @@ The Studio's footer used to carry the full changelog inline; it now shows one li
 history lives here. Newest first. (Per-session engineering detail stays in
 `arena-studio-handover.md` and the design docs — this file is the user-facing what-changed list.)
 
+## v0.72 (2026-09-04) · Data repos: lab repo picker, "Rig id", no default repo
+
+- **Repo is now a picker.** File ▾ → GitHub → Repo offers the CSHL 2026 course repo,
+  the new private lab repo (`reiserlab/arena-experiments`) and "Other…" for any
+  owner/name. A never-configured browser has **no repo selected** — saves go to local
+  files until one is picked (before, the course repo was silently the default).
+  Course benches already have their repo set, so nothing changes for them.
+- **"Bench id" becomes "Rig id"** outside the course repo, with a pick-list fed by the
+  repo's `roster.yaml` (people's `rig_id`s plus an optional `rigs:` list, which also
+  carries the controller MAC for the connect-time cross-check).
+- **Wording follows the repo** everywhere — save labels ("Save → Reiser lab
+  experiments"), the Open-from-repo picker ("This rig", "Shared protocols
+  (lab-wide)"), promote, run-log push, quick links, help texts. The Pattern Designer
+  and the data-browser dashboard use the same registry.
+- **Expired token handling**: a 401 during a run-log commit now clears the stored
+  token and says so, instead of only "saved locally". Setup guide:
+  `docs/development/data-repo-setup.md`.
+
 ## v0.71 (2026-09-04) · Closed-loop apply is reset at run start and abort
 
 - **No more error floods at the start of a run.** If FicTrac closed-loop "apply"

docs/development/cshl-pipeline-test-plan.md

@@ -137,8 +137,10 @@ while destination/write settings are **locked by default**:
    closes — a classic footgun for a kiosk bench).
 2. Click **🛡 Safe mode**, enter the instructor password, then click the **🔒**
    in the GitHub block to unlock destination/write setup.
-3. **Repo** = `reiserlab/cshl-2026-course`.
-4. **Bench id** = `bench01`…`bench07` (must match `roster.yaml`).
+3. **Repo** = pick *CSHL 2026 course* in the dropdown (Studio v0.72+: nothing is
+   pre-selected on a fresh browser; "Other…" takes a custom owner/name).
+4. **Bench id** = `bench01`…`bench07` (must match `roster.yaml`; the field is
+   labelled "Bench id" while the course repo is selected, "Rig id" otherwise).
 5. Check **"Commit directly to default branch"**.
 6. Click the lock again to **re-lock** (🔒). It re-locks automatically on the
    next page load, so students can't alter the token/repo/bench id.

docs/development/data-repo-rig-test-checklist.md

@@ -0,0 +1,74 @@
+# Data-repo rig test checklist (Studio v0.72, PR #178)
+
+Validates the data-repo generalization on a real rig: repo picker, "Rig id",
+roster-fed rig list, and the full save / open / promote / run-log path against the
+private lab repo `reiserlab/arena-experiments`. About 30 minutes with a rig.
+
+**Before you start.** Serve the PR branch (`git checkout feat/data-repo-registry`,
+`pixi run serve`, open `http://127.0.0.1:8000/arena_studio.html`) or merge #178 and
+hard-refresh the rig after the Pages deploy (Cmd+Shift+R). You need your own
+fine-grained token for `arena-experiments` (`data-repo-token-runbook.md` §B).
+Steps 1–5 and 13–14 need no token and no arena.
+
+## A. Picker and labels (no token)
+
+1. **Fresh state.** Private window → Studio. GitHub block: Repo "— none (local
+   files) —", label **Rig id**, Save button "Save → local file", footer v0.72.
+2. **Lock.** Repo dropdown is greyed. 🛡 advanced → 🔓 enables it; 🔒 greys it again.
+3. **Lab repo.** Pick *Reiser lab experiments*. Label stays **Rig id**, placeholder
+   `e.g. 3e229-g6a`, bottom-corner quick links read "Reiser lab experiments ↗" and
+   point at `reiserlab/arena-experiments`.
+4. **Course repo.** Pick *CSHL 2026 course*. Label flips to **Bench id**, placeholder
+   `e.g. bench03`, quick links repoint. Switch back to the lab repo.
+5. **Other.** Pick *Other…*, type `nonsense`, Enter → banner "Repo must be
+   owner/name", previous repo restored. Type a valid `owner/name` → it sticks.
+
+## B. Signed in against the lab repo
+
+6. **Sign in.** File ▾ → Sign in… → paste your fine-grained token → YES to remember
+   (rig computer). Block reads `✓ @<you> → reiserlab/arena-experiments`. Run-view log
+   shows five "… from the Reiser lab experiments" lines (roster, genotypes, ages,
+   sexes, fly_numbers). Experimenter dropdown = the Janelia names from the roster.
+7. **Rig id.** Click the Rig id field: no suggestions yet (lab roster has no rigs).
+   Type e.g. `3e229-g6a`, tick **Commit directly to default branch**, 🔒.
+8. **Save.** Open any protocol → Save. Button reads "Save → Reiser lab experiments";
+   the file appears under `protocols/<rig-id>/` on GitHub within seconds.
+9. **Open from Repo.** Shows "This rig — <rig-id>" with your save and "Shared
+   protocols (lab-wide)" (empty).
+10. **Promote.** File ▾ → Promote to shared… → file appears in `protocols/shared/`
+    and under the shared header of the picker.
+11. **Roster rigs + MAC chip.** In the repo, edit `roster.yaml`:
+    ```yaml
+    rigs:
+      - rig_id: 3e229-g6a
+        mac: "00:00:00:00:00:00"
+    ```
+    Reload the Studio, connect the arena. The Rig id field now suggests that id; the
+    "⚠ rig ≠ roster" chip appears (wrong MAC). Put the real MAC from the connect log
+    into the roster → reload → chip gone.
+12. **Run log.** Run a short protocol as an *experiment* (bridge connected). Expect
+    "✓ Run log committed" and a file under `runlogs/<rig-id>/`.
+
+## C. Other pages
+
+13. **Pattern Designer.** Open via the Studio's *Patterns ↗* link. ⇪ Save to Repo →
+    the protocol destination row reads "rig <rig-id>" (not "bench").
+14. **Dashboard.** `dashboard/data-browser/`: the repo field suggests both repos; pick
+    the lab repo → it lists your `runlogs/<rig-id>` folder.
+
+## D. Course-bench regression
+
+15. On a bench that already had the course repo stored, load v0.72: Repo shows
+    *CSHL 2026 course* pre-selected, label **Bench id**, "Save → CSHL 2026 course"
+    works as before.
+
+## Known limits to keep in mind
+
+- **Run-log size.** The browser commits via the GitHub Contents API, which rejects
+  files above ~35 MiB (measured; HTTP 422). A 20 s-trial P3 full run is ~26 MB; a
+  40 s-trial run is ~51 MB and will NOT auto-commit — the Studio says "saved locally
+  on the bridge machine". Push those from a clone (`git push` allows up to 100 MB per
+  file) until the Studio gains a large-file path (see release notes / issues).
+- Aborted or test runs never auto-commit (by design) — use ⇪ Push log.
+
+If anything in B misbehaves, the Run-view log line is the fastest diagnostic.

docs/development/data-repo-setup.md

@@ -0,0 +1,61 @@
+# Data repos — setup for a lab rig (and how the course repo fits)
+
+Arena Studio, the Pattern Designer and the data-browser dashboard read and write a
+GitHub **data repo**: protocols (`protocols/<rig-id>/`), their colocated patterns
+(`…_patterns/`), the shared pattern library (`patterns/`), promoted protocols
+(`protocols/shared/`) and one run log per completed recorded run
+(`runlogs/<rig-id>/`). The root YAMLs (`roster.yaml`, `genotypes.yaml`, `ages.yaml`,
+`sexes.yaml`, `fly_numbers.yaml`) are the controlled pick-lists for the Run-details
+panel while that repo is configured; otherwise the site library
+(`configs/metadata/*.yaml`) is used.
+
+Two repos are registered in `js/studio-data-repos.js` and offered in the Studio's
+File ▾ → GitHub **Repo** picker:
+
+| Repo | Use | Visibility | Id label |
+|---|---|---|---|
+| `reiserlab/cshl-2026-course` | CSHL 2026 course showcase (kept as-is) | public | Bench id |
+| `reiserlab/arena-experiments` | the lab's day-to-day experiments | private | Rig id |
+
+A fresh browser has **no repo selected** — saves go to local files until someone picks
+one. "Other…" accepts any `owner/name`.
+
+## Set up a rig computer (once)
+
+1. Get a token — see `data-repo-token-runbook.md` §B (org members: fine-grained,
+   resource owner `reiserlab`, only the repo, Contents read/write).
+2. Arena Studio → **File ▾** → GitHub **Sign in…** → paste → **YES** to remember on a
+   dedicated rig computer (NO on a shared laptop).
+3. 🛡 advanced mode → click **🔓** in the GitHub block → **Repo** = *Reiser lab
+   experiments* → **Rig id** = this station's id (pick from the roster list or type a
+   new one, e.g. `3e229-g6a`; unique per rig) → tick **Commit directly to default
+   branch** → **🔒**.
+4. Add the rig to the repo's `roster.yaml` under `rigs:` (with the controller MAC
+   from the Studio connect log) and give people `rig_id`s if they own a station.
+
+The label next to the id field follows the repo: "Bench id" for the course repo,
+"Rig id" otherwise. The localStorage key is the same (`studio_bench_id`); switching
+repos does not reset it — check it when you switch.
+
+## Sharing across the lab
+
+Everyone with access sees every rig's folders. **File ▾ → Promote to shared…** copies
+a protocol and its `_patterns/` into `protocols/shared/` (refuses to overwrite a
+*different* same-named file). The Pattern Designer's **Save to Repo → library**
+writes `patterns/`. Nothing else is needed.
+
+## Adding another data repo
+
+1. Create + seed it: `scripts/seed-data-repo.sh --repo owner/name --apply`
+   (dry run without `--apply`; private by default; idempotent).
+2. Add an entry to `DATA_REPOS` in `js/studio-data-repos.js` (label, id label,
+   placeholder, shared label, visibility). Tests: `tests/test-studio-data-repos.js`.
+3. Never enable branch protection — the Studio commits straight to the default branch.
+
+## Migrating rig folders between repos
+
+Planned for the week of 2026-09-08 (`scripts/migrate-data-repo-dirs.sh`, see the
+plan): copy `protocols/<id>/` + `runlogs/<id>/` from the course repo into the lab
+repo under a new rig id without rewriting file contents; record the mapping in
+`MIGRATION.md`. Removing from the public repo hides the files but not their git
+history.

docs/development/studio-github-save-proposal.md

@@ -1,6 +1,6 @@
 # Studio → GitHub save: lean proposal (for review)
 
-**Status: COURSE SCOPE SHIPPED (2026-07-03, Session 2 — Arena Studio v0.5).**
+**Status: COURSE SCOPE SHIPPED (2026-07-03, Session 2 — Arena Studio v0.5). PER-LAB SCOPE SHIPPED 2026-09-04 (Studio v0.72) as a repo *registry* (`js/studio-data-repos.js`, private `reiserlab/arena-experiments`, per-person tokens) rather than a template repo — see `data-repo-setup.md`.**
 The course-pipeline slice of this proposal is implemented; the generic
 per-user/template-repo phases below remain the unbuilt follow-on. Shipped:
 repo owner/name + bench-id settings + "commit directly to default branch"