fix(web): address self-review findings on #103 (master-safe defaults + hardening)

Make the branch genuinely master-mergeable and fix several review findings:

- Defaults are now prod-safe: CODEMAN_INSTANCE defaults to '' (→ ~/.codeman,
  -L codeman) and the web port back to 3000, so an existing install upgrades
  cleanly. Port also honors a new CODEMAN_PORT env var. Run the beta isolated
  alongside prod with scripts/run-beta.sh (CODEMAN_INSTANCE=beta + PORT 5000).
- .gitignore: anchor the root `public` symlink rule to `/public` (a bare
  `public` also swallowed src/web/public, silently un-staging new web assets);
  ignore the gesture wasm/model binaries explicitly instead.
- span-displays: add a macOS-only guard (400 elsewhere instead of spawning a
  bash that fails invisibly); extract resolveSpanUrl() for unit testing.
- server.ts: memoize asset-version stat() calls (~1s TTL) so each index render
  doesn't re-stat every script/link tag.
- styles.css: hide the multi-monitor button in solo (detached) windows.
- app.js: require two consecutive unanswered roll-calls before redocking, so a
  timer-throttled background popup isn't wrongly un-marked.
- index.html: make the "skip to terminal" link base-href-safe (onclick scroll)
  so it doesn't navigate to the dashboard from a /session/:id window.
- Tests: test/config/instance.test.ts, test/routes/system-span-displays.test.ts.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

.gitignore

@@ -53,7 +53,17 @@ scripts/remotion/out/
 # Artifacts that should not be tracked
 test-results/
 tmp/
-public
+# Root `public` (a symlink to scripts/remotion/public — local artifact). ANCHORED
+# with a leading slash so it does NOT also match src/web/public (a bare `public`
+# would swallow the whole web UI source dir and silently un-stage any new asset
+# added there). No trailing slash so it still matches the symlink, not just dirs.
+/public
+
+# Opt-in gesture overlay runtime assets: large MediaPipe wasm + model (~27 MB)
+# fetched at build/install by scripts/fetch-gesture-assets.mjs, kept out of git.
+# (The gesture bundle itself, gesture-codeman.js, IS tracked.)
+src/web/public/gesture/wasm/
+src/web/public/gesture/*.task
 
 # Claude Code plan tracking
 plan.json

CLAUDE.md

@@ -30,7 +30,7 @@ This file provides guidance to Claude Code (claude.ai/code) when working with co
 2. **Frontend changes**: Use Playwright to load the page and assert the UI renders correctly. Use `waitUntil: 'domcontentloaded'` (not `networkidle` — SSE keeps the connection open). Wait 3-4s for polling/async data to populate, then check element visibility, text content, and CSS values
 3. **Only after verification passes**, proceed with COM
 
-The production server caches static files for 1 year, `immutable` (`maxAge: '1y'` in `server.ts`). To avoid stale frontend after a deploy, `renderIndexHtml` runs `cacheBustAssets(html)` — it appends `?v=<mtime>` to **every same-origin `.js`/`.css`** reference (re-stat'd per render; external/already-versioned/missing refs untouched). Because `index.html` is served `no-cache`, a **normal reload now picks up edited modules/styles — no hard refresh needed** (the gesture bundle is injected separately with its own `?v=`). If you add an asset referenced by an *absolute* URL or from JS rather than a `<script>/<link>` tag, it won't be auto-busted.
+The production server caches static files for 1 year, `immutable` (`maxAge: '1y'` in `server.ts`). To avoid stale frontend after a deploy, `renderIndexHtml` runs `cacheBustAssets(html)` — it appends `?v=<mtime>` to **every same-origin `.js`/`.css`** reference (mtime memoized ~1s so a burst of renders is cheap; external/already-versioned/missing refs untouched). Because `index.html` is served `no-cache`, a **normal reload now picks up edited modules/styles — no hard refresh needed** (the gesture bundle is injected separately with its own `?v=`). If you add an asset referenced by an *absolute* URL or from JS rather than a `<script>/<link>` tag, it won't be auto-busted.
 
 ## COM Shorthand (Deployment)
 
@@ -72,7 +72,7 @@ Codeman is a Claude Code session manager with web interface and autonomous Ralph
 
 ## Additional Commands
 
-`npm run dev` = dev server. Default port: `5000` on this `beta/session-detach` branch (`3000` on master). Commands not in Quick Reference:
+`npm run dev` = dev server. Default port: `3000` (override with `--port` or the `CODEMAN_PORT` env var). To run this beta isolated alongside a prod Codeman, use `scripts/run-beta.sh` (sets `CODEMAN_INSTANCE=beta` + `CODEMAN_PORT=5000`). Commands not in Quick Reference:
 
 | Task | Command |
 |------|---------|
@@ -98,7 +98,7 @@ Codeman is a Claude Code session manager with web interface and autonomous Ralph
 - **Dual-CLI prefix discipline** — Codeman supports both Claude Code and OpenCode (`claude-cli-resolver.ts` / `opencode-cli-resolver.ts`); env-var prefix is CLI-specific (`CLAUDE_CODE_*` vs `OPENCODE_*`) and the allowlist in `schemas.ts` enforces this. When adding settings, decide which CLI(s) it applies to and gate the env export accordingly — don't blindly forward both prefixes. See `docs/opencode-integration.md` for the OpenCode resolver design
 - **Zod `.optional()` rejects `null`** — accepts `undefined` only. When the frontend builds a request body with `JSON.stringify`, an explicit `null` field is preserved on the wire and fails validation with `INVALID_INPUT`. Convert `null` → `undefined` before stringifying (e.g. `field: value ?? undefined`), or declare the schema `.nullish()`. Real bugs caused: 0.6.4 (`durationMinutes` for ∞ respawn), and the same shape pattern hit `opusContext1mEnabled` in 0.6.3
 - **`xterm-zerolag-input` is duplicated** — the local-echo overlay lives in BOTH `packages/xterm-zerolag-input/src/` (published to npm as a standalone library for external consumers — see README "Published Packages") AND inline inside `src/web/public/app.js` (runtime copy the web UI actually loads, since the page ships as plain JS without a bundler). Any change to overlay behavior MUST be applied to both, or dev and prod diverge — and a public API break in the package warrants a separate version bump for `xterm-zerolag-input` in the changeset. Always test on mobile after touching it. See `docs/local-echo-overlay-plan.md`.
-- **Instance isolation / multi-instance attach danger** — data dir (`~/.codeman`) and tmux socket (`tmux -L codeman`) are PROCESS-WIDE and shared by every Codeman on the machine, derived from `CODEMAN_INSTANCE` via `src/config/instance.ts` (`getDataDir()`/`dataPath()`/`DEFAULT_TMUX_SOCKET`). ⚠️ A 2nd instance on the SAME socket **discovers and attaches PTYs to the first instance's live sessions** (`tmux -L codeman attach-session …`), resizing/mutating them — `$HOME` isolation is NOT enough (tmux is system-global). To run two instances, give each a distinct `CODEMAN_INSTANCE` (scopes BOTH dir+socket: `~/.codeman-<name>` + `-L codeman-<name>`), or set `CODEMAN_TMUX_SOCKET` + `CODEMAN_DATA_DIR` individually. **This `beta/session-detach` branch defaults to `CODEMAN_INSTANCE=beta` and port 5000** so it coexists with a prod Codeman out of the box. Any new `~/.codeman/...` path MUST go through `dataPath()`, never `join(homedir(), '.codeman', …)`.
+- **Instance isolation / multi-instance attach danger** — data dir (`~/.codeman`) and tmux socket (`tmux -L codeman`) are PROCESS-WIDE and shared by every Codeman on the machine, derived from `CODEMAN_INSTANCE` via `src/config/instance.ts` (`getDataDir()`/`dataPath()`/`DEFAULT_TMUX_SOCKET`). ⚠️ A 2nd instance on the SAME socket **discovers and attaches PTYs to the first instance's live sessions** (`tmux -L codeman attach-session …`), resizing/mutating them — `$HOME` isolation is NOT enough (tmux is system-global). To run two instances, give each a distinct `CODEMAN_INSTANCE` (scopes BOTH dir+socket: `~/.codeman-<name>` + `-L codeman-<name>`), or set `CODEMAN_TMUX_SOCKET` + `CODEMAN_DATA_DIR` individually. **`CODEMAN_INSTANCE` defaults to empty = the production layout (`~/.codeman`, `-L codeman`, port 3000)**, so this branch is safe to ship to master without disturbing existing installs. To run THIS beta alongside prod, launch with `scripts/run-beta.sh` (`CODEMAN_INSTANCE=beta` + `CODEMAN_PORT=5000`) — it never collides with prod's data dir/socket/port. Any new `~/.codeman/...` path MUST go through `dataPath()`, never `join(homedir(), '.codeman', …)`.
 
 **Import conventions**: Utils from `./utils`, types from `./types` (barrel), config from specific `./config/*` files.
 

scripts/fetch-gesture-assets.mjs

@@ -3,8 +3,8 @@
  * gesture-recognizer model) into src/web/public/gesture/ so Codeman can serve
  * them same-origin (a browser content-blocker otherwise blocks the public CDNs
  * and the overlay fails to start). These are large binaries (~27 MB) kept OUT of
- * git (the bare `public` rule in .gitignore covers them); they are fetched here
- * at install (postinstall) and build time instead.
+ * git (ignored explicitly via `src/web/public/gesture/wasm/` + `*.task` in
+ * .gitignore); they are fetched here at install (postinstall) and build time.
  *
  * Idempotent: skips files already present. Non-fatal: the gesture overlay is
  * opt-in (CODEMAN_GESTURE=1), so a fetch failure only warns — it must not break

scripts/run-beta.sh

@@ -0,0 +1,32 @@
+#!/usr/bin/env bash
+#
+# run-beta.sh — launch a BETA Codeman isolated from a production instance.
+#
+# Codeman's data dir (~/.codeman) and tmux socket (-L codeman) are process-wide
+# and shared by every instance on the machine. The code now DEFAULTS to that
+# production layout on port 3000 (safe for master / existing installs), so a beta
+# build no longer isolates itself automatically — this wrapper opts it in:
+#
+#   CODEMAN_INSTANCE=beta  → data dir ~/.codeman-beta + tmux socket codeman-beta
+#   CODEMAN_PORT=5000      → listen on 5000 instead of 3000
+#
+# Result: the beta runs side-by-side with prod and can never discover/attach to
+# prod's live tmux sessions or clobber prod's state.json. Override either var to
+# run additional named instances, e.g. CODEMAN_INSTANCE=foo CODEMAN_PORT=5050.
+#
+# Usage:  ./scripts/run-beta.sh [extra `codeman web` flags]
+# Build first (the beta runs the compiled dist):  npm run build
+
+set -euo pipefail
+
+export CODEMAN_INSTANCE="${CODEMAN_INSTANCE:-beta}"
+export CODEMAN_PORT="${CODEMAN_PORT:-5000}"
+
+DIST="$(cd "$(dirname "$0")/.." && pwd)/dist/index.js"
+if [ ! -f "$DIST" ]; then
+  echo "dist not found at $DIST — run 'npm run build' first." >&2
+  exit 1
+fi
+
+echo "Starting beta Codeman: instance='$CODEMAN_INSTANCE' (~/.codeman-$CODEMAN_INSTANCE, -L codeman-$CODEMAN_INSTANCE) on port $CODEMAN_PORT"
+exec node "$DIST" web "$@"

src/cli.ts

@@ -483,7 +483,7 @@ program
 program
   .command('web')
   .description('Start the web interface')
-  .option('-p, --port <port>', 'Port to listen on', '5000')
+  .option('-p, --port <port>', 'Port to listen on (env: CODEMAN_PORT)', process.env.CODEMAN_PORT || '3000')
   .option('--https', 'Enable HTTPS with self-signed certificate (only needed for remote access, not localhost)')
   .option('--title-hostname <hostname>', 'Override the hostname shown in the browser title')
   .action(async (options) => {

src/config/instance.ts

@@ -9,10 +9,16 @@
  *
  * To let a beta build coexist with a production one, this module derives both
  * the data dir and the tmux socket from a single "instance" name:
- *   - default (this branch): `beta` → `~/.codeman-beta` + `tmux -L codeman-beta`
- *   - `CODEMAN_INSTANCE=`     (empty) → `~/.codeman`     + `tmux -L codeman`   (prod layout)
+ *   - default (unset/empty) → `~/.codeman`      + `tmux -L codeman`        (prod layout)
+ *   - `CODEMAN_INSTANCE=beta` → `~/.codeman-beta` + `tmux -L codeman-beta`
  *   - `CODEMAN_INSTANCE=foo`  → `~/.codeman-foo`  + `tmux -L codeman-foo`
  *
+ * The DEFAULT is the production layout so this is safe to ship to master: an
+ * existing install keeps reading `~/.codeman`. To run a beta ALONGSIDE prod,
+ * launch it with `CODEMAN_INSTANCE=beta` (and a distinct port, see below) —
+ * `scripts/run-beta.sh` does both. The port is unrelated to the instance and is
+ * set separately via `--port` / `CODEMAN_PORT` (see `src/cli.ts`).
+ *
  * Individual overrides still win: `CODEMAN_DATA_DIR` (absolute data dir) and
  * `CODEMAN_TMUX_SOCKET` (socket name, validated in tmux-manager).
  */
@@ -22,12 +28,12 @@ import { join } from 'node:path';
 import { mkdirSync } from 'node:fs';
 
 /**
- * Instance name. Empty string = production layout (`~/.codeman`, `-L codeman`).
- * Defaults to `beta` on the beta/session-detach branch so it never collides
- * with a production Codeman. Set `CODEMAN_INSTANCE=` (empty) to opt back into
- * the production layout.
+ * Instance name. Empty string (the default) = production layout (`~/.codeman`,
+ * `-L codeman`), so this is safe on master and existing installs are untouched.
+ * Set `CODEMAN_INSTANCE=beta` (e.g. via `scripts/run-beta.sh`) to run an
+ * isolated beta alongside prod.
  */
-export const CODEMAN_INSTANCE = process.env.CODEMAN_INSTANCE ?? 'beta';
+export const CODEMAN_INSTANCE = process.env.CODEMAN_INSTANCE ?? '';
 
 const INSTANCE_SUFFIX = CODEMAN_INSTANCE ? `-${CODEMAN_INSTANCE}` : '';
 

src/web/public/app.js

@@ -309,6 +309,7 @@ class CodemanApp {
     this._redockGrace = new Map();       // id -> timer: deferred redock (debounces popup reloads)
     this._detachPingPending = null;      // Set of ids awaiting a liveness answer
     this._detachLivenessTimer = null;    // periodic reconcile of channel-only detached windows
+    this._detachOrphanStrikes = new Map(); // id -> consecutive unanswered roll-calls (redock at 2)
 
     this._initGeneration = 0;     // dedup concurrent handleInit calls
     this._initFallbackTimer = null; // fallback timer if SSE init doesn't arrive
@@ -879,6 +880,7 @@ class CodemanApp {
     const t = this._detachWatchTimers.get(id);
     if (t) { clearInterval(t); this._detachWatchTimers.delete(id); }
     this._cancelPendingRedock(id);
+    this._detachOrphanStrikes.delete(id);
     this.detachedWindows.delete(id);
     this._markDetached(id, false);
   }
@@ -965,6 +967,7 @@ class CodemanApp {
     if (msg.type === 'detached' && msg.id) {
       this._cancelPendingRedock(msg.id);    // a re-announce (e.g. popup reload) cancels a deferred redock
       this._detachPingPending?.delete(msg.id);  // and proves liveness for this tick
+      this._detachOrphanStrikes.delete(msg.id); // any answer clears accumulated misses
       this._markDetached(msg.id, true);
     } else if (msg.type === 'redocked' && msg.id) {
       this._scheduleRedock(msg.id);         // defer: a popup reload fires redocked→detached; grace avoids a badge blip
@@ -993,10 +996,18 @@ class CodemanApp {
     if (!orphans.length) return;
     this._detachPingPending = new Set(orphans);
     this._postWindowMessage({ type: 'roll-call' });
-    // Live popups answer 'detached' (clearing themselves above); survivors are gone.
+    // Live popups answer 'detached' (clearing themselves above); survivors stay in
+    // the pending set. Redock only after TWO consecutive unanswered roll-calls — a
+    // backgrounded popup is timer-throttled and may miss a single 1.2s window, and
+    // we don't want to wrongly un-mark a still-open tab. A later answer resets the
+    // strike count (see _onWindowMessage).
     setTimeout(() => {
       if (!this._detachPingPending) return;
-      for (const id of this._detachPingPending) this._redock(id);
+      for (const id of this._detachPingPending) {
+        const strikes = (this._detachOrphanStrikes.get(id) || 0) + 1;
+        if (strikes >= 2) { this._detachOrphanStrikes.delete(id); this._redock(id); }
+        else this._detachOrphanStrikes.set(id, strikes);
+      }
       this._detachPingPending = null;
     }, 1200);
   }

src/web/public/index.html

@@ -59,7 +59,9 @@
     <div class="skeleton-toolbar"></div>
   </div>
   <!-- Skip link for keyboard users -->
-  <a href="#terminalContainer" class="skip-link">Skip to terminal</a>
+  <!-- onclick scrolls/focuses directly: with <base href="/"> a bare href="#..." would
+       navigate to /#... (the dashboard) from a /session/:id solo window. -->
+  <a href="#terminalContainer" class="skip-link" onclick="event.preventDefault(); var t=document.getElementById('terminalContainer'); if(t){t.scrollIntoView(); var f=t.querySelector('textarea,[tabindex]'); (f||t).focus&&(f||t).focus();}">Skip to terminal</a>
   <div class="app">
     <!-- Compact Header with Session Tabs -->
     <header class="header">

src/web/public/styles.css

@@ -963,6 +963,7 @@ body.solo-mode .session-tabs,
 body.solo-mode .header-system-stats,
 body.solo-mode .header-tokens,
 body.solo-mode .btn-notifications,
+body.solo-mode .btn-multimonitor,
 body.solo-mode .btn-lifecycle-log {
   display: none !important;
 }

src/web/routes/system-routes.ts

@@ -94,6 +94,18 @@ function getSystemStats(): {
   }
 }
 
+/**
+ * Build the URL the spanning browser window should open, pinned to localhost.
+ * Takes only a digits-only port from the (untrusted) Host header so nothing
+ * attacker-controllable reaches the launched browser; falls back to the default
+ * port when the header is absent/odd. Exported for unit testing.
+ */
+export function resolveSpanUrl(hostHeader: string | undefined, fallbackPort = '3000'): string {
+  const hostPort = String(hostHeader ?? '').split(':')[1] ?? '';
+  const port = /^\d+$/.test(hostPort) ? hostPort : fallbackPort;
+  return `http://localhost:${port}`;
+}
+
 export function registerSystemRoutes(
   app: FastifyInstance,
   ctx: SessionPort & EventPort & ConfigPort & InfraPort & AuthPort
@@ -250,17 +262,21 @@ export function registerSystemRoutes(
   // panels can be dragged across the physical monitor seam. macOS only; needs
   // the one-time "Displays have separate Spaces" OFF prerequisite (see script).
   app.post('/api/system/span-displays', async (req, reply) => {
+    // macOS only: the launcher uses osascript + Finder desktop bounds and Chrome
+    // --app geometry flags. Fail clearly elsewhere instead of spawning a bash
+    // that errors out invisibly (the toast would otherwise lie "Opening…").
+    if (process.platform !== 'darwin') {
+      return reply
+        .code(400)
+        .send(createErrorResponse(ApiErrorCode.INVALID_INPUT, 'Multi-monitor spanning is only supported on macOS.'));
+    }
     // Resolve the bundled launcher relative to this module (works from src/ and dist/).
     const scriptPath = join(dirname(fileURLToPath(import.meta.url)), '../../../scripts/span-codeman.sh');
     if (!existsSync(scriptPath)) {
       return reply.code(500).send(createErrorResponse(ApiErrorCode.INTERNAL_ERROR, 'span-codeman.sh not found'));
     }
-    // Point the spanning window at THIS server. Pin the host to localhost (same
-    // machine) and take only a digits-only port from the Host header so nothing
-    // attacker-controllable reaches the launched browser.
-    const hostPort = String(req.headers.host ?? '').split(':')[1] ?? '';
-    const port = /^\d+$/.test(hostPort) ? hostPort : '5000';
-    const url = `http://localhost:${port}`;
+    // Point the spanning window at THIS server (localhost + sanitized port).
+    const url = resolveSpanUrl(req.headers.host);
     try {
       const child = spawn('bash', [scriptPath, url], { detached: true, stdio: 'ignore' });
       child.on('error', (err) => app.log.error({ err }, 'span-displays launch failed'));

src/web/server.ts

@@ -1021,37 +1021,47 @@ export class WebServer extends EventEmitter {
     return html;
   }
 
-  /** Cache-busting query for the gesture bundle: its mtime, re-read per render.
-   *  The bundle is served from /gesture/ with a 1-year cache, so without a
-   *  version that changes on redeploy the browser would keep running a stale
-   *  bundle forever. Re-stat'ing each render means a freshly copied-in bundle is
-   *  picked up with no server restart. Empty string if the file is missing. */
-  private gestureBundleVersion(): string {
+  /** mtime memo for asset cache-busting (keyed by absolute path). A full index
+   *  render does one stat per script/link tag (~25-30); without this each `/`,
+   *  `/index.html` and `/session/:id` hit would re-stat them all. A 1s TTL keeps
+   *  a burst of renders cheap while still picking up an edited/redeployed file
+   *  within a second (no server restart needed). */
+  private _assetVersionMemo = new Map<string, { v: number; ts: number }>();
+  private assetVersion(absPath: string): number | null {
+    const now = Date.now();
+    const hit = this._assetVersionMemo.get(absPath);
+    if (hit && now - hit.ts < 1000) return hit.v;
     try {
-      const p = join(__dirname, 'public', 'gesture', 'gesture-codeman.js');
-      return `?v=${Math.floor(statSync(p).mtimeMs)}`;
+      const v = Math.floor(statSync(absPath).mtimeMs);
+      this._assetVersionMemo.set(absPath, { v, ts: now });
+      return v;
     } catch {
-      return '';
+      return null;
     }
   }
 
+  /** Cache-busting query for the gesture bundle: its mtime (memoized, see
+   *  assetVersion). The bundle is served from /gesture/ with a 1-year cache, so
+   *  without a version that changes on redeploy the browser would keep running a
+   *  stale bundle forever. Empty string if the file is missing. */
+  private gestureBundleVersion(): string {
+    const v = this.assetVersion(join(__dirname, 'public', 'gesture', 'gesture-codeman.js'));
+    return v === null ? '' : `?v=${v}`;
+  }
+
   /** Append ?v=<mtime> to every same-origin .js/.css reference in the page so a
    *  normal reload always serves the latest. Codeman's static assets are sent
    *  with `Cache-Control: max-age=1y, immutable` and the script/link tags carry
    *  no version, so without this an edited module (panels-ui.js, styles.css, …)
-   *  stays cached until a manual hard refresh. mtime is re-stat'd per render, so
-   *  a changed file is picked up with no server restart. External URLs (have a
+   *  stays cached until a manual hard refresh. mtime is memoized (1s TTL) so a
+   *  changed file is picked up with no server restart. External URLs (have a
    *  `:` scheme), already-versioned refs (have a `?`), and refs with no matching
    *  file on disk are left untouched. */
   private cacheBustAssets(html: string): string {
     const publicDir = join(__dirname, 'public');
     return html.replace(/(\s(?:src|href)=")([^"?:]+\.(?:js|css))(")/g, (full, pre, ref, post) => {
-      try {
-        const v = Math.floor(statSync(join(publicDir, ref)).mtimeMs);
-        return `${pre}${ref}?v=${v}${post}`;
-      } catch {
-        return full;
-      }
+      const v = this.assetVersion(join(publicDir, ref));
+      return v === null ? full : `${pre}${ref}?v=${v}${post}`;
     });
   }
 

test/config/instance.test.ts

@@ -0,0 +1,79 @@
+/**
+ * Per-instance isolation (src/config/instance.ts): the data dir + tmux socket
+ * derive from CODEMAN_INSTANCE, defaulting to the production layout so the
+ * feature branch is safe to merge to master.
+ *
+ * instance.ts reads env at module load, so each case re-imports it via
+ * vi.resetModules() under a controlled env. node:fs mkdirSync is mocked so
+ * getDataDir() never creates real directories on the test machine.
+ *
+ * Port: N/A (no server).
+ */
+import { describe, it, expect, afterEach, vi } from 'vitest';
+import { homedir } from 'node:os';
+import { join } from 'node:path';
+
+vi.mock('node:fs', async (orig) => {
+  const actual = await orig<typeof import('node:fs')>();
+  return { ...actual, mkdirSync: vi.fn() };
+});
+
+const ENV_KEYS = ['CODEMAN_INSTANCE', 'CODEMAN_DATA_DIR'] as const;
+const ORIG: Record<string, string | undefined> = Object.fromEntries(ENV_KEYS.map((k) => [k, process.env[k]]));
+
+async function load(env: Partial<Record<(typeof ENV_KEYS)[number], string | undefined>> = {}) {
+  vi.resetModules();
+  for (const k of ENV_KEYS) {
+    const v = env[k];
+    if (v === undefined) delete process.env[k];
+    else process.env[k] = v;
+  }
+  return import('../../src/config/instance.js');
+}
+
+afterEach(() => {
+  for (const k of ENV_KEYS) {
+    if (ORIG[k] === undefined) delete process.env[k];
+    else process.env[k] = ORIG[k];
+  }
+  vi.resetModules();
+});
+
+describe('config/instance', () => {
+  it('defaults to the production layout when CODEMAN_INSTANCE is unset', async () => {
+    const m = await load({ CODEMAN_INSTANCE: undefined, CODEMAN_DATA_DIR: undefined });
+    expect(m.CODEMAN_INSTANCE).toBe('');
+    expect(m.DEFAULT_TMUX_SOCKET).toBe('codeman');
+    expect(m.getDataDir()).toBe(join(homedir(), '.codeman'));
+    expect(m.dataPath('state.json')).toBe(join(homedir(), '.codeman', 'state.json'));
+  });
+
+  it('treats an explicitly-empty CODEMAN_INSTANCE as the production layout', async () => {
+    const m = await load({ CODEMAN_INSTANCE: '', CODEMAN_DATA_DIR: undefined });
+    expect(m.CODEMAN_INSTANCE).toBe('');
+    expect(m.DEFAULT_TMUX_SOCKET).toBe('codeman');
+    expect(m.getDataDir()).toBe(join(homedir(), '.codeman'));
+  });
+
+  it('scopes BOTH the data dir and the tmux socket for a named instance', async () => {
+    const m = await load({ CODEMAN_INSTANCE: 'beta', CODEMAN_DATA_DIR: undefined });
+    expect(m.CODEMAN_INSTANCE).toBe('beta');
+    expect(m.DEFAULT_TMUX_SOCKET).toBe('codeman-beta');
+    expect(m.getDataDir()).toBe(join(homedir(), '.codeman-beta'));
+    expect(m.dataPath('mux-sessions.json')).toBe(join(homedir(), '.codeman-beta', 'mux-sessions.json'));
+  });
+
+  it('supports an arbitrary instance name', async () => {
+    const m = await load({ CODEMAN_INSTANCE: 'foo', CODEMAN_DATA_DIR: undefined });
+    expect(m.DEFAULT_TMUX_SOCKET).toBe('codeman-foo');
+    expect(m.getDataDir()).toBe(join(homedir(), '.codeman-foo'));
+  });
+
+  it('CODEMAN_DATA_DIR overrides the derived data dir (socket still instance-scoped)', async () => {
+    const m = await load({ CODEMAN_INSTANCE: 'beta', CODEMAN_DATA_DIR: '/tmp/codeman-test-xyz' });
+    expect(m.getDataDir()).toBe('/tmp/codeman-test-xyz');
+    expect(m.dataPath('a', 'b')).toBe(join('/tmp/codeman-test-xyz', 'a', 'b'));
+    // Socket is derived from the instance name, not the data dir override.
+    expect(m.DEFAULT_TMUX_SOCKET).toBe('codeman-beta');
+  });
+});

test/routes/system-span-displays.test.ts

@@ -0,0 +1,76 @@
+/**
+ * POST /api/system/span-displays (multi-monitor launcher) + resolveSpanUrl.
+ *
+ * The route shells out to scripts/span-codeman.sh, so we mock child_process.spawn
+ * to avoid actually opening a browser (and to assert the sanitized URL passed to
+ * it). process.platform is overridden per-case so the macOS-only guard is tested
+ * deterministically regardless of where the suite runs.
+ *
+ * Port: N/A (app.inject).
+ */
+import { describe, it, expect, afterEach, vi } from 'vitest';
+
+const spawnMock = vi.hoisted(() => vi.fn(() => ({ on: vi.fn(), unref: vi.fn() })));
+vi.mock('node:child_process', async (orig) => {
+  const actual = await orig<typeof import('node:child_process')>();
+  return { ...actual, spawn: spawnMock };
+});
+
+import { createRouteTestHarness } from './_route-test-utils.js';
+import { registerSystemRoutes, resolveSpanUrl } from '../../src/web/routes/system-routes.js';
+
+const REAL_PLATFORM = process.platform;
+function setPlatform(p: NodeJS.Platform) {
+  Object.defineProperty(process, 'platform', { value: p, configurable: true });
+}
+afterEach(() => {
+  setPlatform(REAL_PLATFORM);
+  spawnMock.mockClear();
+});
+
+describe('resolveSpanUrl', () => {
+  it('takes a digits-only port from the Host header, pinned to localhost', () => {
+    expect(resolveSpanUrl('localhost:5000')).toBe('http://localhost:5000');
+    // Hostname is discarded — always localhost (same machine).
+    expect(resolveSpanUrl('attacker.example.com:3000')).toBe('http://localhost:3000');
+  });
+
+  it('falls back to the default port for missing / non-numeric ports', () => {
+    expect(resolveSpanUrl(undefined)).toBe('http://localhost:3000');
+    expect(resolveSpanUrl('localhost')).toBe('http://localhost:3000');
+    expect(resolveSpanUrl('localhost:99;rm -rf /')).toBe('http://localhost:3000');
+    expect(resolveSpanUrl('localhost:80abc')).toBe('http://localhost:3000');
+    expect(resolveSpanUrl('x', '5000')).toBe('http://localhost:5000');
+  });
+});
+
+describe('POST /api/system/span-displays', () => {
+  it('returns 400 (macOS-only) on non-darwin and never spawns', async () => {
+    setPlatform('linux');
+    const { app } = await createRouteTestHarness(registerSystemRoutes);
+    const res = await app.inject({ method: 'POST', url: '/api/system/span-displays' });
+    expect(res.statusCode).toBe(400);
+    expect(res.json().success).toBe(false);
+    expect(res.json().error).toMatch(/macOS/i);
+    expect(spawnMock).not.toHaveBeenCalled();
+    await app.close();
+  });
+
+  it('spawns the launcher with the sanitized localhost URL on darwin', async () => {
+    setPlatform('darwin');
+    const { app } = await createRouteTestHarness(registerSystemRoutes);
+    const res = await app.inject({
+      method: 'POST',
+      url: '/api/system/span-displays',
+      headers: { host: 'localhost:5000' },
+    });
+    expect(res.statusCode).toBe(200);
+    expect(res.json()).toMatchObject({ success: true, url: 'http://localhost:5000' });
+    expect(spawnMock).toHaveBeenCalledTimes(1);
+    const [cmd, args] = spawnMock.mock.calls[0] as [string, string[]];
+    expect(cmd).toBe('bash');
+    expect(args[0]).toMatch(/span-codeman\.sh$/);
+    expect(args[1]).toBe('http://localhost:5000');
+    await app.close();
+  });
+});