token-harness 0.1.12 → 0.1.14

package/README.md

@@ -29,90 +29,67 @@ of CLI commands to memorize.
 
 ### First run
 
-1. Open **Dashboard** and let Token Harness inspect the current setup.
-2. If setup is incomplete, choose **Open setup**.
-3. In **Setup**, work from top to bottom:
-   - Coding agents
-   - Managed optimizers
-   - Optional agent tuning
-   - Checks and maintenance
-4. For Claude Code or Codex, choose **Review baseline** when the RTK + HarnessTrim setup is available.
-5. Read the exact proposed changes. **Apply reviewed setup** appears only when there is a concrete
-   safe plan to apply.
-6. Keep using Claude Code or Codex normally.
-7. Open **Results** when you want to see what Token Harness can actually prove.
+The first screen is **Overview**. There is no separate Setup page to learn.
 
-Opening the app does not change your configuration. Read-only checks stay read-only, and a managed
-write requires an explicit review and approval.
+1. Token Harness detects Claude Code and Codex.
+2. Each detected coding agent says either **Ready** or **Setup incomplete**.
+3. If setup is incomplete, choose **Finish setup** on that agent. Token Harness prepares the
+   recommended **RTK + HarnessTrim** baseline and shows the exact safe plan before anything changes.
+4. The **Optimizers** section keeps the recommended baseline separate from optional tools
+   (mcptoon, GitNexus and Headroom). When a tool needs an action, that action appears on the same
+   card as the status it refers to.
+5. **Health and updates** is maintenance, not another onboarding checklist. Normal setup performs its
+   own safety checks. Use **Re-check health** for troubleshooting and **Check for updates** when you
+   want to inspect provider versions. If a reviewed update is available, the same dialog offers
+   **Install updates** after showing the versions.
+6. Keep using Claude Code or Codex normally. Open **Results** when you want detailed evidence.
 
-## The three views
+Opening the app does not change your configuration. A configuration or software change is always
+previewed first and requires an explicit review and approval.
 
-### Dashboard
+## The two views
 
-Dashboard answers the questions that matter first:
+### Overview
 
-- is my setup ready;
-- which coding agents and managed optimizers are active;
-- what should I do next;
-- what value has actually been measured;
-- whether quality or an integration needs attention.
+Overview is both the first-run screen and the normal status screen. It keeps the two product entities
+separate:
 
-The headline cards deliberately distinguish measured evidence from unknown values. Missing evidence
-is never shown as zero savings.
+- **Coding agents** — currently Claude Code and Codex. Detection only means Token Harness can see the
+  agent; first-run setup is complete for that agent only when both RTK and HarnessTrim are connected.
+- **Optimizers** — RTK and HarnessTrim are the recommended baseline. mcptoon, GitNexus and Headroom
+  are optional reviewed integrations with narrower prerequisites and compatibility boundaries.
 
-### Setup
+The status at the top always answers what to do next. States such as **Setup incomplete**,
+**Installed · not connected**, **Needs attention**, or **Update available** have their action beside
+the affected agent or optimizer instead of in a separate action list.
 
-Setup is one ordered workflow instead of a collection of unrelated actions.
+Overview also contains a compact **Measured impact** summary. Missing evidence is shown as unknown,
+never as zero savings.
 
-**1. Coding agents**
+Advanced agent details and reasoning preferences are collapsed because they are not required for
+first-run optimizer setup.
 
-Token Harness currently supports guided setup for Claude Code and Codex. Agent details also show
-useful read-only allowance and connected-tool observations when available.
+### Optimizer lifecycle and evidence
 
-**2. Managed optimizers**
+The recommended baseline remains **RTK + HarnessTrim** on individually reviewed combinations.
+Token Harness also exposes **mcptoon**, **GitNexus** and **Headroom** as optional managed integrations
+on exact reviewed lifecycle rows. Enabling an optional optimizer does not make it part of the
+production baseline and does not create a savings claim.
 
-The production baseline remains **RTK + HarnessTrim** on individually reviewed combinations.
-Token Harness also exposes **mcptoon**, **GitNexus** and **Headroom** as optional managed integrations on their exact
-reviewed lifecycle rows; enabling any of them does not make it part of the production baseline or create a
-savings claim. Token Harness tracks the exact combined provider set separately: if no combined-stack
-review is recorded, Setup says so and keeps the stack incomplete rather than inferring compatibility
-from healthy individual checks. Token Harness can prepare their integration transactionally, show the
-exact plan, apply it only after approval, verify it, and remove only configuration it owns.
+Token Harness can prepare supported integration changes transactionally, show the exact plan, apply
+it only after approval, verify what the declared tier can verify, and remove only configuration it
+owns. GitNexus is never auto-indexed and its noncommercial license boundary stays visible. Headroom
+remains config-only managed on its reviewed row; Token Harness does not bootstrap uv/Python or start
+wrapper/proxy/deploy flows.
 
-For maintainers validating the combined stack, `token-harness stack-review --json` captures the exact
-configured provider versions and managed harness sets and reuses the existing passive `verify`
-evidence for exact provider/harness pairs. Runtime evidence is credited only when it can be
-attributed to that harness: HarnessTrim uses its native event harness field, while provider-wide
-telemetry can be attributed by exclusion only when one harness is wired. With multiple harnesses, an
-unattributable receipt stays **Unavailable** instead of being copied across rows. A `not-exercised`
-result does not promise that ordinary agent use will create a receipt: the provider must record a
-qualifying operation, and for reducers that means a real reduction. `stack-review` does not run an
-active canary or spend a model call, and it never makes the compatibility decision itself. The
-shipped combined-review registry stays empty until the captured configuration has enough real
-verification evidence, has been benchmarked together, and has been deliberately reviewed; see
-`docs/combined-stack-reviews.md`.
+For maintainers, `token-harness stack-review --json` captures the exact configured provider versions
+and managed harness sets for combined-stack review. Runtime evidence is credited only when it can be
+attributed to the relevant harness; missing attribution remains unavailable rather than being copied
+across rows.
 
-**3. Evaluation evidence (maintainers)**
-
-Selection campaigns and historical candidate evidence remain available for maintainers, but they are
-not a separate novice setup product. Once a tool has a reviewed managed lifecycle, Setup presents it in
-the same **Optimization Stack** with its exact state, version, prerequisites, harness scope and safe
-Apply/Remove controls. Evaluation evidence never turns into a savings or compatibility claim by itself.
-
-Headroom is config-only managed on its exact reviewed 0.37.0 row. Token Harness can own the narrow
-Claude Code or Codex MCP registration after the pinned CLI is present; it deliberately does not
-bootstrap `uv`/Python, start wrappers/proxies, or claim runtime compression savings. GitNexus likewise
-does not auto-index repositories, and its noncommercial license boundary stays visible.
-
-**4. Optional agent tuning**
-
-Reasoning preferences are separate from optimizer installation. They are persistent agent settings,
-not hidden per-task switches, and are changed only through the normal preview/apply flow.
-
-**5. Checks and maintenance**
-
-Read-only integration checks, update checks and safe removal/undo controls live here. An update
-outside reviewed compatibility is not forced.
+Evaluation campaigns are an advanced maintainer workflow, not a novice setup step. Normal managed
+setup stays in the unified **Optimization Stack** and historical candidate evidence never turns into
+a savings, compatibility or promotion claim by itself.
 
 ### Results
 
@@ -384,8 +361,9 @@ Update Token Harness itself:
 npm install --global token-harness@latest
 ```
 
-Provider checks and reviewed updates are in **Setup -> Checks and maintenance**. From the advanced
-CLI, first preview and then explicitly apply the provider-package updates:
+Provider checks and reviewed updates are in **Overview -> Health and updates**. The browser first
+checks reviewed channels and, when an installable update exists, offers **Install updates** in the
+same dialog. From the advanced CLI, the preview prints the exact confirmation command:
 
 ```sh
 token-harness update
@@ -442,7 +420,7 @@ Node must be at least 22.13. Reopen the terminal after installation if needed.
 
 ### Setup or verification needs attention
 
-Use the action shown in Dashboard or Setup. For technical evidence:
+Use the action shown beside the affected agent or optimizer in Overview. For technical evidence:
 
 ```sh
 token-harness doctor --verbose

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "token-harness",
-  "version": "0.1.12",
+  "version": "0.1.14",
   "description": "Quota-aware efficiency layer for Claude Code and Codex subscription limits.",
   "license": "Apache-2.0",
   "type": "module",

package/sbom.json

@@ -1,14 +1,14 @@
 {
   "bomFormat": "CycloneDX",
   "specVersion": "1.5",
-  "serialNumber": "urn:uuid:584dee19-a861-2f17-f82a-93efadc2de0b",
+  "serialNumber": "urn:uuid:8840bbd7-fcea-b78f-b033-a40bef562512",
   "version": 1,
   "metadata": {
     "component": {
       "type": "application",
       "bom-ref": "token-harness",
       "name": "token-harness",
-      "version": "0.1.12",
+      "version": "0.1.14",
       "description": "Quota-aware efficiency layer for Claude Code and Codex subscription limits.",
       "licenses": [
         {
@@ -20,7 +20,7 @@
       "hashes": [
         {
           "alg": "SHA-256",
-          "content": "584dee19a8612f17f82a93efadc2de0b6c8b28fc96bd7a7d44b9e5af2dbb09bc"
+          "content": "8840bbd7fceab78fb033a40bef5625123772648c48cb4fea73b39020b714e9cf"
         }
       ]
     },

package/token-harness.mjs

@@ -9387,7 +9387,7 @@ var TOOL_VERSION;
 var init_version2 = __esm({
   "apps/cli/dist/src/version.js"() {
     "use strict";
-    TOOL_VERSION = "0.1.12";
+    TOOL_VERSION = "0.1.14";
   }
 });
 
@@ -28522,7 +28522,7 @@ async function runUpdate(context) {
         receiptId: null
       }
     };
-    return result5(EXIT_CODES["confirmation-required"], null, [
+    return result5(EXIT_CODES["confirmation-required"], report2, [
       ...preparedDiagnostics,
       adoptionConfirmationDiagnostic(plan6),
       confirmationDiagnostic(report2)
@@ -32753,30 +32753,43 @@ var init_guided = __esm({
           let network = false;
           for (const agent of selected) {
             this.record(`Preparing supported changes for ${name(agent.harnessId)}. No settings changed.`, "working");
-            const args = ["plan", "--harness", agent.harnessId];
-            if (data["action"] === "setup" && data["provider"] !== void 0)
-              args.push("--provider", String(data["provider"]));
-            if (data["action"] === "skill") {
-              args.push("--provider", "none", "--agent-skill");
-            } else if (data["action"] === "effort")
-              args.push("--provider", "none", "--native-policy", "--task", String(data["task"]), "--profile", data["task"] === "mechanical" ? "economy" : data["task"] === "standard" ? "balanced" : "quality");
-            const result6 = await this.call(args);
-            const report = result6.data;
-            if (report === null || result6.exitCode !== 0 || report.conflicts.length > 0 || report.actions.length === 0 || !report.persisted || report.planId === null) {
-              notices.push(`${name(agent.harnessId)}: ${explainGuideIssue(result6.diagnostics, data["action"] === "effort" ? "No supported preference change is needed or available. Your current preference is kept." : data["action"] === "skill" ? "In-session guidance is already present, or an existing user-owned skill location was left untouched." : "No safe setup change is available. The integration may already be configured, or a required provider is not installed.")}`);
-              continue;
-            }
-            plans.push(report.planId);
-            if (data["action"] === "skill") {
-              changes.push({
-                title: `${name(agent.harnessId)}: enable in-session guidance`,
-                description: "Installs one reviewed Token Harness Agent Skill in the standard user skill directory. It does not change model, login, billing, hooks, trust, or the current conversation.",
-                files: 1
-              });
+            const setupProviders = [];
+            if (data["action"] === "setup" && data["harness"] !== void 0) {
+              if (data["provider"] === void 0) {
+                setupProviders.push("rtk", "harnesstrim");
+              } else {
+                setupProviders.push(String(data["provider"]));
+              }
             } else {
-              changes.push(...report.actions.map((action2) => describeChange(action2, agent.harnessId)));
+              setupProviders.push(null);
+            }
+            for (const setupProvider of setupProviders) {
+              const args = ["plan", "--harness", agent.harnessId];
+              if (setupProvider !== null)
+                args.push("--provider", setupProvider);
+              if (data["action"] === "skill") {
+                args.push("--provider", "none", "--agent-skill");
+              } else if (data["action"] === "effort")
+                args.push("--provider", "none", "--native-policy", "--task", String(data["task"]), "--profile", data["task"] === "mechanical" ? "economy" : data["task"] === "standard" ? "balanced" : "quality");
+              const result6 = await this.call(args);
+              const report = result6.data;
+              if (report === null || result6.exitCode !== 0 || report.conflicts.length > 0 || report.actions.length === 0 || !report.persisted || report.planId === null) {
+                const subject = setupProvider !== null ? `${name(agent.harnessId)} \xB7 ${name(setupProvider)}` : name(agent.harnessId);
+                notices.push(`${subject}: ${explainGuideIssue(result6.diagnostics, data["action"] === "effort" ? "No supported preference change is needed or available. Your current preference is kept." : data["action"] === "skill" ? "In-session guidance is already present, or an existing user-owned skill location was left untouched." : "No safe setup change is available. The integration may already be configured, or a required provider is not installed.")}`);
+                continue;
+              }
+              plans.push(report.planId);
+              if (data["action"] === "skill") {
+                changes.push({
+                  title: `${name(agent.harnessId)}: enable in-session guidance`,
+                  description: "Installs one reviewed Token Harness Agent Skill in the standard user skill directory. It does not change model, login, billing, hooks, trust, or the current conversation.",
+                  files: 1
+                });
+              } else {
+                changes.push(...report.actions.map((action2) => describeChange(action2, agent.harnessId)));
+              }
+              network ||= report.network.length > 0;
             }
-            network ||= report.network.length > 0;
           }
           if (selected.length === 0)
             notices.push("No supported coding agent was found. Install and sign in to Claude Code or Codex, then refresh.");
@@ -32816,6 +32829,52 @@ var init_guided = __esm({
           if (approval === null || approval.id !== data["ticket"] || this.now() >= approval.expires)
             throw new GuideError(409, "This preview expired or was already used. Review a fresh preview.");
           this.approval = null;
+          if (approval.operation === "update") {
+            this.record("Installing only the reviewed provider updates you approved.", "working");
+            let result6;
+            try {
+              result6 = await this.call(["update", "--yes"]);
+            } catch {
+              this.invalidateObservedState();
+              const message = "The update stopped before its final result could be read. No automatic retry was made. Refresh and inspect the installed optimizer versions before retrying.";
+              this.record(message, "attention");
+              return {
+                ok: false,
+                title: "Update result needs checking",
+                messages: [message],
+                appliedPlans: 0
+              };
+            }
+            if (result6.exitCode !== 0 || result6.data === null) {
+              const message = explainGuideIssue(result6.diagnostics, "The reviewed optimizer update was not applied. The installed version, update channel or compatibility evidence changed after the preview, so nothing was forced.");
+              this.invalidateObservedState();
+              this.record(message, "attention");
+              return {
+                ok: false,
+                title: "Optimizer update was not applied",
+                messages: [message],
+                appliedPlans: 0
+              };
+            }
+            const applied = result6.data.execution?.results.filter((row2) => row2.status === "applied") ?? [];
+            this.lastApplied = null;
+            this.invalidateObservedState();
+            const messages2 = applied.length > 0 ? [
+              "The reviewed optimizer update completed successfully.",
+              "Token Harness re-ran the update transaction safety checks before changing software.",
+              "Choose Refresh to read the installed versions again. Reopen a coding agent if the updated optimizer requires it."
+            ] : [
+              "No optimizer needed an update by the time the approved action ran.",
+              "Nothing was forced. Choose Refresh to read the current versions."
+            ];
+            this.record(applied.length > 0 ? "Reviewed optimizer update installed." : "Optimizer update no longer needed.", "success");
+            return {
+              ok: true,
+              title: applied.length > 0 ? "Optimizer updated" : "Already up to date",
+              messages: messages2,
+              appliedPlans: applied.length
+            };
+          }
           if (approval.operation === "candidate-apply" || approval.operation === "candidate-uninstall") {
             if (approval.candidate === void 0 || approval.candidateHarness === void 0)
               throw new GuideError(409, "This experimental preview is incomplete. Review it again.");
@@ -32971,6 +33030,7 @@ var init_guided = __esm({
       }
       async checkUpdates() {
         return this.exclusive(async () => {
+          this.approval = null;
           this.record("Checking provider update channels without changing software.", "working");
           const updateResult = await this.call(["update"]);
           const inventory = await this.call(["doctor"]);
@@ -32982,7 +33042,8 @@ var init_guided = __esm({
               ok: false,
               title: "Update check needs attention",
               messages: [message],
-              appliedPlans: 0
+              appliedPlans: 0,
+              ticket: null
             };
           }
           const fingerprint = stackFingerprint(inventory.data);
@@ -33004,7 +33065,8 @@ var init_guided = __esm({
               title: "Update check needs attention",
               messages: [message],
               appliedPlans: 0,
-              stack: stack2
+              stack: stack2,
+              ticket: null
             };
           }
           this.updates = { fingerprint, report: updateResult.data };
@@ -33012,19 +33074,43 @@ var init_guided = __esm({
           const blocked = updateResult.data.providers.filter((row2) => row2.verdict === "blocked-unreviewed");
           const messages = [];
           if (available.length > 0) {
-            messages.push(...available.map((row2) => `${name(row2.providerId)}: ${row2.installed ?? "installed version"} \u2192 ${row2.available ?? "new version"} is available. Run token-harness update to review the dry-run; applying still requires explicit approval.`));
+            messages.push(...available.map((row2) => `${name(row2.providerId)}: ${row2.installed ?? "installed version"} \u2192 ${row2.available ?? "new version"} is reviewed and ready to install.`));
           }
           if (blocked.length > 0) {
-            messages.push(...blocked.map((row2) => `${name(row2.providerId)}: ${row2.available ?? "a newer version"} exists, but Token Harness is keeping ${row2.installed ?? "the installed version"} until that combination has reviewed compatibility evidence.`));
+            messages.push(...blocked.map((row2) => `${name(row2.providerId)}: ${row2.available ?? "a newer version"} exists, but Token Harness is keeping ${row2.installed ?? "the installed version"} because that newer combination is not reviewed yet.`));
           }
           if (available.length === 0 && blocked.length === 0)
-            messages.push("No reviewed provider update is currently available. Pins and unavailable channels remain visible in the stack.");
-          messages.push("This was a read-only channel check. No provider was installed, updated, downgraded or enabled.");
+            messages.push("Your managed optimizers are up to date on their reviewed channels.");
+          let ticket = null;
+          if (available.length > 0) {
+            ticket = this.random();
+            const expires = this.now() + 10 * 6e4;
+            this.approval = {
+              id: ticket,
+              expires,
+              plans: [],
+              transactionId: null,
+              provider: null,
+              description: "Optimizer updates",
+              operation: "update",
+              network: updateResult.data.network.length > 0
+            };
+            messages.push("Review the versions above, then choose Install updates. Token Harness will re-check the update channels and transaction safety rules before changing software.");
+          } else {
+            messages.push("No software was changed.");
+          }
           const stack = this.stackSnapshot();
           if (this.cached !== null)
             this.cached.value = { ...this.cached.value, stack };
-          this.record("Provider update check completed. No software changed.", "success");
-          return { ok: true, title: "Optimizer update check", messages, appliedPlans: 0, stack };
+          this.record(available.length > 0 ? "Reviewed optimizer update available. Waiting for your approval." : "Provider update check completed. No reviewed update is pending.", "success");
+          return {
+            ok: true,
+            title: available.length > 0 ? "Updates available" : "Optimizers up to date",
+            messages,
+            appliedPlans: 0,
+            stack,
+            ticket
+          };
         });
       }
       async verify() {
@@ -34767,9 +34853,20 @@ var init_guided_capabilities_client = __esm({
     return data;
   }
 
-  function markStale() {
-    $('stale-state').hidden = false;
-    $('stale-state').textContent = 'A reviewed change was applied. Displayed status is the previous state until you choose Refresh.';
+  async function waitForAutomaticRefresh() {
+    $('stale-state').hidden = true;
+    $('stale-state').textContent = '';
+    const spinner = $('reading-spinner');
+    const refresh = $('refresh');
+    let observedRefresh = false;
+    refresh.click();
+    const deadline = Date.now() + 120000;
+    while (Date.now() < deadline) {
+      if (!spinner.hidden) observedRefresh = true;
+      if (observedRefresh && spinner.hidden) return;
+      await new Promise(resolve => setTimeout(resolve, 100));
+    }
+    throw new Error('The latest setup state could not be read automatically.');
   }
 
   async function applyTicket(ticket, title) {
@@ -34782,11 +34879,33 @@ var init_guided_capabilities_client = __esm({
       const result = await request('/api/apply', { ticket });
       if (activeRun !== runId) return;
       $('modal-title').textContent = result.ok ? title : 'Change needs attention';
-      $('modal-content').replaceChildren(message(result.ok ? 'Completed' : 'Needs attention', (result.messages || []).join(' ') || 'The transaction finished.', result.ok ? 'safe' : 'warn'));
-      if (result.ok) markStale();
+      $('modal-content').replaceChildren(
+        message(
+          result.ok ? 'Applied' : 'Needs attention',
+          (result.messages || []).join(' ') || 'The transaction finished.',
+          result.ok ? 'safe' : 'warn',
+        ),
+        progress(
+          'Refreshing the current setup',
+          'Please wait while Token Harness re-reads the latest agent and optimizer state.',
+        ),
+      );
+      $('modal-actions').replaceChildren(closeButton('Refreshing…', true));
+      await waitForAutomaticRefresh();
+      if (activeRun !== runId) return;
+      $('modal-content').replaceChildren(
+        message(
+          result.ok ? 'Completed and refreshed' : 'Action finished; status refreshed',
+          (result.messages || []).join(' ') || 'The latest setup state is now displayed.',
+          result.ok ? 'safe' : 'warn',
+        ),
+      );
       $('modal-actions').replaceChildren(closeButton('Done'));
     } catch (error) {
       if (activeRun !== runId) return;
+      $('stale-state').hidden = false;
+      $('stale-state').textContent =
+        'The action finished, but Token Harness could not automatically refresh the latest setup state.';
       $('modal-error').textContent = error.message;
       $('modal-error').hidden = false;
       $('modal-actions').replaceChildren(closeButton('Close'));
@@ -34808,14 +34927,22 @@ var init_guided_capabilities_client = __esm({
       const data = await request('/api/preview', body);
       if (activeRun !== runId || !$('modal').open) return;
       $('modal-content').replaceChildren();
-      if (!(data.changes || []).length)
-        $('modal-content').append(message('No managed change proposed', (data.notices || []).join(' ') || 'Nothing needs to be changed.'));
-      for (const change of data.changes || []) {
-        const item = node('article', undefined, 'preview-change');
-        item.append(node('h3', change.title), node('p', change.description));
-        $('modal-content').append(item);
-      }
-      for (const notice of data.notices || []) $('modal-content').append(node('p', notice, 'notice-row'));
+      if (!(data.changes || []).length) {
+        $('modal-content').append(
+          message(
+            'No managed change proposed',
+            (data.notices || []).join(' ') || 'Nothing needs to be changed.',
+          ),
+        );
+      } else {
+        for (const change of data.changes || []) {
+          const item = node('article', undefined, 'preview-change');
+          item.append(node('h3', change.title), node('p', change.description));
+          $('modal-content').append(item);
+        }
+        for (const notice of data.notices || [])
+          $('modal-content').append(node('p', notice, 'notice-row'));
+      }
       $('modal-content').append(message('Safety', 'Only Token Harness-owned changes can be applied or removed. User-owned provider installations and unrelated configuration are left untouched.', 'safe'));
       $('modal-actions').replaceChildren(closeButton(data.ticket ? 'Cancel' : 'Done'));
       if (data.ticket)
@@ -35024,9 +35151,8 @@ var init_guided_product_client = __esm({
   const count = value => new Intl.NumberFormat(undefined, { maximumFractionDigits: 1 }).format(value);
   const date = value => value ? new Date(value).toLocaleString() : 'not recorded';
   const VIEWS = {
-    dashboard: ['Dashboard', 'Your coding setup, what is active, and what to do next.'],
-    setup: ['Setup', 'Install and configure optimization tools in a clear order.'],
-    results: ['Results', 'Measured savings, quality evidence, and recent activity.'],
+    dashboard: ['Overview', 'Your coding agents, optimizer setup, health and measured results in one place.'],
+    results: ['Results', 'Detailed measured savings, quality evidence, and recent activity.'],
   };
   const TOOL_INFO = {
     rtk: {
@@ -35232,6 +35358,24 @@ var init_guided_product_client = __esm({
     return (current?.stack?.components || []).filter(component => component.configured);
   }
 
+  function componentConfiguredFor(id, agentId) {
+    const component = managedComponent(id);
+    return Boolean(component?.configuredHarnesses?.includes(agentId));
+  }
+
+  function baselineReadyFor(agentId) {
+    return componentConfiguredFor('rtk', agentId) && componentConfiguredFor('harnesstrim', agentId);
+  }
+
+  function baselineIncompleteAgents() {
+    return activeAgents().filter(agent => !baselineReadyFor(agent.id));
+  }
+
+  function providerSupportsAgent(providerId, agentId) {
+    if (providerId === 'gitnexus') return agentId === 'claude';
+    return agentId === 'claude' || agentId === 'codex';
+  }
+
   function candidateObservation(id) {
     return (current?.optimizationCandidates || []).find(item => item.id === id) || null;
   }
@@ -35278,21 +35422,21 @@ var init_guided_product_client = __esm({
 
   function setupAssessment() {
     const agents = activeAgents();
-    const configured = configuredProviders();
+    const incomplete = baselineIncompleteAgents();
     const stack = current?.stack;
     if (!agents.length)
       return {
         label: 'Agent needed',
         cls: 'warn',
-        title: 'Connect a supported coding agent first',
-        detail: 'Token Harness could not find Claude Code or Codex in the environment that started this app.',
-        action: 'setup',
+        title: 'Install or expose Claude Code or Codex first',
+        detail: 'Token Harness could not find a supported coding agent in the environment that started this app.',
+        action: 'none',
       };
     if (current?.value?.quality?.state === 'regressed')
       return {
         label: 'Needs attention',
         cls: 'warn',
-        title: 'Review the quality regression before increasing optimization',
+        title: 'Measured quality needs attention',
         detail: 'A paired benchmark favored the baseline. Token Harness is not crediting the affected savings.',
         action: 'results',
       };
@@ -35300,31 +35444,33 @@ var init_guided_product_client = __esm({
       return {
         label: 'Needs attention',
         cls: 'warn',
-        title: 'One integration needs attention',
-        detail: 'Open Setup to see the affected tool and run the read-only integration check.',
-        action: 'setup',
+        title: 'One configured optimizer needs attention',
+        detail: 'Use Re-check health below to see whether the configured integration still matches its reviewed state.',
+        action: 'verify',
       };
-    if (!configured.length)
+    if (incomplete.length)
       return {
         label: 'Setup incomplete',
         cls: 'warn',
-        title: 'Your coding agent is ready; add the managed optimization stack',
-        detail: 'Start with RTK and HarnessTrim. You will review every proposed configuration change before it is applied.',
-        action: 'setup',
+        title: incomplete.length === 1
+          ? 'Finish setup for ' + incomplete[0].name
+          : 'Finish the recommended setup for your coding agents',
+        detail: 'The recommended baseline is RTK + HarnessTrim. Finish setup on each incomplete coding-agent card below; you will see the exact safe plan before anything changes.',
+        action: 'configure',
       };
     if (!(current?.savings?.rows || []).length)
       return {
         label: 'Ready',
         cls: 'good',
-        title: 'Setup is ready; use your coding agent normally',
-        detail: 'Token Harness has no measured result yet. Keep working normally and results will appear after evidence is recorded.',
+        title: 'Recommended setup is complete',
+        detail: 'Use your coding agents normally. Measured results will appear when Token Harness has evidence to report.',
         action: 'results',
       };
     return {
       label: 'Ready',
       cls: 'good',
-      title: 'Your managed setup is active',
-      detail: 'No urgent action is required. Review measured results or manage optional optimizers when you want to.',
+      title: 'Your recommended setup is active',
+      detail: 'No setup action is required. The summary below shows what Token Harness can actually measure.',
       action: 'results',
     };
   }
@@ -35342,15 +35488,14 @@ var init_guided_product_client = __esm({
     const assessment = setupAssessment();
     $('dashboard-status').replaceChildren();
     const main = node('div');
-    main.append(node('span', 'SETUP STATUS', 'eyebrow'), node('h2', assessment.title), node('p', assessment.detail));
+    main.append(node('span', 'CURRENT STATUS', 'eyebrow'), node('h2', assessment.title), node('p', assessment.detail));
     const actions = node('div', undefined, 'inline-actions');
-    actions.append(
-      assessment.action === 'setup'
-        ? navigateButton('Open setup', 'setup')
-        : navigateButton('View results', 'results'),
-    );
-    if (assessment.action !== 'setup') actions.append(navigateButton('Manage setup', 'setup', 'secondary'));
-    main.append(actions);
+    if (assessment.action === 'verify') {
+      actions.append(actionButton('Re-check health', () => readOnlyOperation('verify')));
+    } else if (assessment.action === 'results') {
+      actions.append(navigateButton('View detailed results', 'results'));
+    }
+    if (actions.children.length) main.append(actions);
     $('dashboard-status').append(main, pill(assessment.label, assessment.cls));
 
     const reduction = bestReduction();
@@ -35364,46 +35509,18 @@ var init_guided_product_client = __esm({
       metricCard('API cost', 'Not measured yet', 'Requires billed-token evidence and a verified price basis.'),
       metricCard('Quality', quality.value, quality.detail, quality.cls),
     );
-
-    const active = $('dashboard-active');
-    active.replaceChildren();
-    for (const agent of activeAgents()) {
-      const row = node('div', undefined, 'status-row');
-      row.append(node('div', agent.name + (agent.version ? ' · v' + agent.version : '')), pill(agent.state, agent.configured ? 'good' : ''));
-      active.append(row);
-    }
-    for (const component of configuredProviders()) {
-      const row = node('div', undefined, 'status-row');
-      const name = TOOL_INFO[component.providerId]?.name || component.displayName || component.providerId;
-      const configuredFor = (component.configuredHarnesses || []).map(agentName).join(', ');
-      row.append(node('div', name + (configuredFor ? ' · ' + configuredFor : '')), pill(component.health === 'healthy' ? 'Verified' : 'Configured', component.health === 'attention' ? 'warn' : 'good'));
-      active.append(row);
-    }
-    if (!active.children.length) active.append(sectionEmpty('No managed optimization is active yet. Open Setup to start.'));
-
-    const checklist = $('dashboard-checklist');
-    checklist.replaceChildren();
-    const agentReady = activeAgents().length > 0;
-    const managedReady = configuredProviders().length > 0;
-    const evidenceReady = (current?.savings?.rows || []).length > 0;
-    for (const item of [
-      [agentReady, 'Coding agent detected', agentReady ? activeAgents().map(agent => agent.name).join(', ') : 'Claude Code or Codex must be visible to this process.'],
-      [managedReady, 'Managed optimizer configured', managedReady ? configuredProviders().map(item => TOOL_INFO[item.providerId]?.name || item.providerId).join(', ') : 'RTK/HarnessTrim setup has not been completed yet.'],
-      [evidenceReady, 'Measured result available', evidenceReady ? 'Recorded optimization evidence is available.' : 'Use your configured coding agent normally to build evidence.'],
-    ]) {
-      const row = node('div', undefined, 'check-row');
-      row.append(node('span', item[0] ? '✓' : '○', 'check-icon ' + (item[0] ? 'done' : '')), node('div'));
-      row.lastChild.append(node('strong', item[1]), node('p', item[2], 'caption'));
-      checklist.append(row);
-    }
   }
 
   function componentState(component) {
-    if (!component) return { label: 'Not detected', cls: '' };
+    if (!component || component.detectedState === 'absent') return { label: 'Not installed', cls: '' };
     if (component.health === 'attention') return { label: 'Needs attention', cls: 'warn' };
-    if (component.configured) return { label: component.verification === 'verified' ? 'Configured + verified' : 'Configured', cls: 'good' };
-    if (component.installed) return { label: 'Installed · setup needed', cls: '' };
-    return { label: 'Not installed/configured', cls: '' };
+    if (component.configured)
+      return {
+        label: component.verification === 'verified' ? 'Connected · verified' : 'Connected',
+        cls: 'good',
+      };
+    if (component.installed) return { label: 'Installed · not connected', cls: '' };
+    return { label: 'Not installed', cls: '' };
   }
 
   function renderManagedTool(id) {
@@ -35415,29 +35532,69 @@ var init_guided_product_client = __esm({
     const title = node('div');
     title.append(
       node('h3', info.name),
-      node('span', info.optional ? 'Optional managed integration' : 'Production baseline', 'caption'),
+      node('span', info.optional ? 'Optional optimizer' : 'Recommended baseline', 'caption'),
     );
     head.append(title, pill(state.label, state.cls));
     card.append(head, node('p', info.role));
+
     const facts = node('div', undefined, 'tool-facts');
+    if (component?.version) facts.append(node('span', 'Installed version'), node('strong', 'v' + component.version));
     const configuredFor = component?.configuredHarnesses?.length
       ? component.configuredHarnesses.map(agentName).join(', ')
-      : 'No agent yet';
-    facts.append(node('span', 'Configured for'), node('strong', configuredFor));
-    if (component?.version) facts.append(node('span', 'Version'), node('strong', 'v' + component.version));
+      : 'Not connected to a coding agent';
+    facts.append(node('span', 'Connection'), node('strong', configuredFor));
+    if (component?.update === 'available')
+      facts.append(
+        node('span', 'Update'),
+        node('strong', 'v' + (component.updateAvailableVersion || 'new version') + ' available'),
+      );
+    else if (component?.update === 'blocked')
+      facts.append(node('span', 'Update'), node('strong', 'Newer version not reviewed yet'));
     card.append(facts);
+
     if (component?.warnings?.length) {
       const attention = component.health === 'attention';
       const prerequisite = !component.installed && component.nextAction?.kind === 'install-configure';
       if (attention || prerequisite)
         card.append(
           messageBox(
-            attention ? 'Needs attention' : 'Prerequisite needed',
+            attention ? 'Needs attention' : 'Setup requirement',
             component.warnings[0].message,
             attention ? 'warn' : '',
           ),
         );
     }
+
+    const actions = node('div', undefined, 'inline-actions');
+    for (const agent of activeAgents()) {
+      if (!providerSupportsAgent(id, agent.id)) continue;
+      if (component?.configuredHarnesses?.includes(agent.id)) continue;
+      if (info.optional) {
+        actions.append(
+          actionButton(
+            'Set up for ' + agent.name,
+            () => reviewSetup(agent.id, id),
+            'secondary',
+          ),
+        );
+      } else {
+        actions.append(
+          actionButton(
+            component?.installed ? 'Connect to ' + agent.name : 'Set up for ' + agent.name,
+            () => reviewSetup(agent.id, id),
+            'secondary',
+          ),
+        );
+      }
+    }
+    if (component?.update === 'available')
+      actions.append(actionButton('Install update', () => readOnlyOperation('updates'), 'secondary'));
+    if (
+      component?.configured &&
+      (component.health === 'attention' || component.verification === 'degraded')
+    )
+      actions.append(actionButton('Re-check health', () => readOnlyOperation('verify'), 'secondary'));
+    if (actions.children.length) card.append(actions);
     return card;
   }
 
@@ -35445,16 +35602,47 @@ var init_guided_product_client = __esm({
     const root = $('setup-agents');
     root.replaceChildren();
     if (!activeAgents().length) {
-      root.append(messageBox('No supported agent detected', 'Start Token Harness from a terminal where Claude Code or Codex is installed and available on PATH, then choose Refresh.', 'warn'));
+      root.append(
+        messageBox(
+          'No supported coding agent detected',
+          'Start Token Harness from a terminal where Claude Code or Codex is installed and available on PATH, then choose Refresh.',
+          'warn',
+        ),
+      );
       return;
     }
     for (const agent of activeAgents()) {
+      const ready = baselineReadyFor(agent.id);
       const card = node('article', undefined, 'tool-card compact');
       const head = node('div', undefined, 'tool-head');
       const title = node('div');
-      title.append(node('h3', agent.name), node('span', agent.version ? 'v' + agent.version : 'Version unavailable', 'caption'));
-      head.append(title, pill('Detected', 'good'));
-      card.append(head, node('p', agent.configured ? 'At least one managed optimizer is configured for this agent.' : 'The agent is available, but managed optimizer setup is not complete.'));
+      title.append(
+        node('h3', agent.name),
+        node('span', agent.version ? 'v' + agent.version : 'Version unavailable', 'caption'),
+      );
+      head.append(title, pill(ready ? 'Ready' : 'Setup incomplete', ready ? 'good' : 'warn'));
+      card.append(
+        head,
+        node(
+          'p',
+          ready
+            ? 'RTK + HarnessTrim are configured for this coding agent.'
+            : 'The agent is detected. Finish the recommended RTK + HarnessTrim setup to complete onboarding.',
+        ),
+      );
+      if (agent.providers?.length)
+        card.append(
+          node(
+            'p',
+            'Connected optimizers: ' + agent.providers.join(', '),
+            'caption',
+          ),
+        );
+      if (!ready) {
+        const actions = node('div', undefined, 'inline-actions');
+        actions.append(actionButton('Finish setup', () => reviewSetup(agent.id)));
+        card.append(actions);
+      }
       root.append(card);
     }
   }
@@ -35466,16 +35654,18 @@ var init_guided_product_client = __esm({
     const provider = providerId ? TOOL_INFO[providerId] : null;
     if (providerId && !provider) return;
     const run = modal(
-      provider ? 'Review ' + provider.name + ' for ' + agent.name : 'Review managed setup for ' + agent.name,
+      provider
+        ? 'Set up ' + provider.name + ' for ' + agent.name
+        : 'Set up recommended optimizers for ' + agent.name,
     );
     $('modal-content').append(
       messageBox(
-        'What this checks',
+        'Safe preview first',
         provider
-          ? 'Token Harness will inspect only ' + provider.name + ' for ' + agent.name + ' and propose a change only on an exact reviewed compatibility row.'
-          : 'RTK and HarnessTrim are the production baseline. Token Harness will inspect both for ' + agent.name + ' and propose only compatible changes.',
+          ? 'Token Harness will inspect only ' + provider.name + ' for ' + agent.name + '. If setup is supported, the exact change appears before you confirm it.'
+          : 'Token Harness will prepare the recommended RTK + HarnessTrim setup for ' + agent.name + '. The exact files and changes appear before you confirm them.',
       ),
-      messageBox('Nothing changes yet', 'This step is read-only. If a change is available, you will see exactly what it writes before an Apply button appears.'),
+      messageBox('Nothing changes yet', 'This first step only prepares the safe plan. You decide whether to apply it after seeing the concrete changes.'),
       progress('Checking ' + agent.name, 'Reading installed optimizer versions and current integration state.'),
     );
     if (providerId === 'gitnexus')
@@ -35502,22 +35692,34 @@ var init_guided_product_client = __esm({
         $('modal-content').replaceChildren();
         $('modal-content').append(
           messageBox(
-            'Managed scope',
+            provider ? 'This optimizer only' : 'Recommended setup',
             provider
-              ? provider.name + ' is reviewed separately from the RTK + HarnessTrim production baseline. This approval changes only the provider-scoped plan shown below.'
-              : 'This review covers the RTK + HarnessTrim production baseline only. Optional managed integrations are reviewed separately.',
+              ? 'This approval changes only ' + provider.name + ' for ' + agent.name + '. Other optimizers are left unchanged.'
+              : 'This approval covers only RTK + HarnessTrim for ' + agent.name + '. Optional optimizers stay separate.',
           ),
         );
-        if (!data.changes.length) $('modal-content').append(messageBox('No managed change proposed', (data.notices || []).join(' ') || 'This agent already has the supported setup, or no safe managed change is available.'));
-        for (const change of data.changes) {
-          const item = node('article', undefined, 'preview-change');
-          item.append(node('h3', change.title), node('p', change.description));
-          $('modal-content').append(item);
+        if (!data.changes.length) {
+          $('modal-content').append(
+            messageBox(
+              'No setup change available',
+              (data.notices || []).join(' ') ||
+                'This setup is already complete, or a prerequisite/version prevents Token Harness from changing it safely.',
+            ),
+          );
+        } else {
+          for (const change of data.changes) {
+            const item = node('article', undefined, 'preview-change');
+            item.append(node('h3', change.title), node('p', change.description));
+            $('modal-content').append(item);
+          }
+          for (const notice of data.notices || []) $('modal-content').append(node('p', notice, 'notice-row'));
         }
-        for (const notice of data.notices || []) $('modal-content').append(node('p', notice, 'notice-row'));
         $('modal-content').append(messageBox('Safety', 'Apply uses the existing transactional engine with backups, compatibility checks, ownership checks and rollback. Unsupported versions are not forced.', 'safe'));
         $('modal-actions').replaceChildren(modalClose(data.ticket ? 'Cancel' : 'Done'));
-        if (data.ticket) $('modal-actions').append(actionButton('Apply reviewed setup', () => applyTicket(data.ticket), ''));
+        if (data.ticket)
+          $('modal-actions').append(
+            actionButton(provider ? 'Set up ' + provider.name : 'Apply recommended setup', () => applyTicket(data.ticket), ''),
+          );
       })
       .catch(error => {
         if (run !== modalRun) return;
@@ -35529,6 +35731,18 @@ var init_guided_product_client = __esm({
       });
   }
 
+  async function refreshOverviewAfterMutation(message = 'Refreshing setup after the approved change…') {
+    setStatus(message, true);
+    $('stale-state').hidden = true;
+    $('stale-state').textContent = '';
+    const period = $('period').value;
+    current = await request('/api/overview?period=' + encodeURIComponent(period) + '&refresh=1');
+    periodCache.set(period, current);
+    render();
+    await loadActivity();
+    setStatus('Updated at ' + new Date(current.generatedAt).toLocaleTimeString() + '.', false);
+  }
+
   async function applyTicket(ticket) {
     if (!ticket || busy) return;
     const run = modalRun;
@@ -35540,12 +35754,39 @@ var init_guided_product_client = __esm({
       const result = await request('/api/apply', { ticket });
       if (run !== modalRun) return;
       $('modal-title').textContent = result.title || (result.ok ? 'Setup completed' : 'Setup needs attention');
-      $('modal-content').replaceChildren(messageBox(result.ok ? 'Completed' : 'Needs attention', (result.messages || []).join(' '), result.ok ? 'safe' : 'warn'));
-      $('modal-content').append(node('p', 'The dashboard is now marked as previous state. Choose Refresh when you want to re-read the complete setup.', 'caption'));
-      $('stale-state').hidden = false;
-      $('stale-state').textContent = 'A setup change was applied. Displayed status is the previous state until you choose Refresh.';
+      $('modal-content').replaceChildren(
+        messageBox(result.ok ? 'Applied' : 'Needs attention', (result.messages || []).join(' '), result.ok ? 'safe' : 'warn'),
+        progress('Refreshing the current setup', 'Please wait while Token Harness re-reads agents, optimizer connections and health after the change.'),
+      );
+      $('modal-actions').replaceChildren(modalClose('Refreshing…', true));
+      try {
+        await refreshOverviewAfterMutation('Applying the new configuration and refreshing status…');
+        if (run !== modalRun) return;
+        $('modal-content').replaceChildren(
+          messageBox(
+            result.ok ? 'Completed and refreshed' : 'Action finished; status refreshed',
+            (result.messages || []).join(' ') || 'The latest setup state is now shown on the dashboard.',
+            result.ok ? 'safe' : 'warn',
+          ),
+        );
+      } catch (refreshError) {
+        if (run !== modalRun) return;
+        $('stale-state').hidden = false;
+        $('stale-state').textContent = 'The action finished, but Token Harness could not automatically refresh the latest setup state.';
+        $('modal-content').replaceChildren(
+          messageBox(
+            result.ok ? 'Change applied' : 'Action finished',
+            (result.messages || []).join(' ') || 'The action completed.',
+            result.ok ? 'safe' : 'warn',
+          ),
+          messageBox(
+            'Automatic refresh failed',
+            refreshError.message || 'The latest setup state could not be read automatically. Use Refresh to retry the status check.',
+            'warn',
+          ),
+        );
+      }
       $('modal-actions').replaceChildren(modalClose('Done'));
-      loadActivity();
     } catch (error) {
       if (run !== modalRun) return;
       $('modal-error').textContent = error.message;
@@ -35560,45 +35801,15 @@ var init_guided_product_client = __esm({
     $('managed-tools').replaceChildren(
       renderManagedTool('rtk'),
       renderManagedTool('harnesstrim'),
-      renderManagedTool('mcptoon'),
-      renderManagedTool('gitnexus'),
-      renderManagedTool('headroom'),
     );
-    const actions = $('managed-setup-actions');
-    actions.replaceChildren();
-    if (!activeAgents().length) {
-      actions.append(sectionEmpty('A supported coding agent must be detected before managed setup can be reviewed.'));
-      return;
-    }
-    actions.append(node('p', 'Choose the agent you want to configure. Review setup is read-only; Apply appears only after a concrete safe plan is shown.', 'caption'));
-    const buttons = node('div', undefined, 'inline-actions');
-    for (const agent of activeAgents())
-      buttons.append(actionButton('Review baseline for ' + agent.name, () => reviewSetup(agent.id)));
-    actions.append(buttons);
-
-    const optional = node('div', undefined, 'managed-optional');
-    optional.append(
-      node(
-        'p',
-        'Optional managed integrations are reviewed one at a time. Enabling one does not create a savings claim or make it part of the RTK + HarnessTrim production baseline.',
-        'caption',
-      ),
-    );
-    const optionalButtons = node('div', undefined, 'inline-actions');
-    for (const agent of activeAgents()) {
-      optionalButtons.append(
-        actionButton('Review mcptoon for ' + agent.name, () => reviewSetup(agent.id, 'mcptoon'), 'secondary'),
-      );
-      if (agent.id === 'claude')
-        optionalButtons.append(
-          actionButton('Review GitNexus for ' + agent.name, () => reviewSetup(agent.id, 'gitnexus'), 'secondary'),
-        );
-      optionalButtons.append(
-        actionButton('Review Headroom for ' + agent.name, () => reviewSetup(agent.id, 'headroom'), 'secondary'),
+    const optionalRoot = $('optional-tools');
+    if (optionalRoot)
+      optionalRoot.replaceChildren(
+        renderManagedTool('mcptoon'),
+        renderManagedTool('gitnexus'),
+        renderManagedTool('headroom'),
       );
-    }
-    optional.append(optionalButtons);
-    actions.append(optional);
+
   }
 
   function candidateState(observation) {
@@ -35844,19 +36055,44 @@ var init_guided_product_client = __esm({
   function readOnlyOperation(kind) {
     if (busy) return;
     const isVerify = kind === 'verify';
-    const run = modal(isVerify ? 'Check integrations' : 'Check optimizer updates');
+    const run = modal(isVerify ? 'Re-check optimizer health' : 'Check for optimizer updates');
     $('modal-content').append(
-      messageBox('Read-only check', isVerify ? 'Verifies the currently configured managed integrations. It does not repair or change configuration.' : 'Checks update channels for installed managed optimizers. It does not download or upgrade anything.'),
-      progress(isVerify ? 'Checking integrations' : 'Checking update channels', 'Nothing will be changed.'),
+      messageBox(
+        isVerify ? 'Troubleshooting check' : 'Update preview',
+        isVerify
+          ? 'Re-checks the configured optimizer integrations without repairing or changing them. Normal setup already performs its own safety checks.'
+          : 'Checks reviewed update channels first. If an installable update exists, you can approve it from this window.',
+      ),
+      progress(
+        isVerify ? 'Checking configured optimizers' : 'Checking update channels',
+        'Nothing changes during this check.',
+      ),
     );
     $('modal-actions').append(modalClose('Cancel'));
     setBusy(true, false);
     ensureSession()
-      .then(() => request(isVerify ? '/api/verify' : '/api/update-check', { period: $('period').value }))
+      .then(() =>
+        request(isVerify ? '/api/verify' : '/api/update-check', { period: $('period').value }),
+      )
       .then(result => {
         if (run !== modalRun || !$('modal').open) return;
-        $('modal-content').replaceChildren(messageBox(result.ok ? 'Completed' : 'Needs attention', (result.messages || []).join(' ') || 'Check completed.', result.ok ? 'safe' : 'warn'));
-        $('modal-actions').replaceChildren(modalClose('Done'));
+        if (result.stack && current) {
+          current = { ...current, stack: result.stack };
+          renderDashboard();
+          renderSetup();
+        }
+        $('modal-content').replaceChildren(
+          messageBox(
+            result.title || (result.ok ? 'Completed' : 'Needs attention'),
+            (result.messages || []).join(' ') || 'Check completed.',
+            result.ok ? 'safe' : 'warn',
+          ),
+        );
+        $('modal-actions').replaceChildren(modalClose(result.ticket ? 'Cancel' : 'Done'));
+        if (result.ticket)
+          $('modal-actions').append(
+            actionButton('Install updates', () => applyTicket(result.ticket), ''),
+          );
       })
       .catch(error => {
         if (run !== modalRun) return;
@@ -35902,19 +36138,57 @@ var init_guided_product_client = __esm({
   function renderMaintenance() {
     const root = $('maintenance-actions');
     root.replaceChildren();
-    const verify = node('article', undefined, 'maintenance-row');
-    const verifyText = node('div');
-    verifyText.append(node('strong', 'Check integrations'), node('p', 'Read-only verification of configured managed tools.', 'caption'));
-    verify.append(verifyText, actionButton('Check', () => readOnlyOperation('verify'), 'secondary'));
+
+    if (configuredProviders().length) {
+      const verify = node('article', undefined, 'maintenance-row');
+      const verifyText = node('div');
+      verifyText.append(
+        node('strong', 'Optimizer health'),
+        node(
+          'p',
+          current?.stack?.state === 'attention'
+            ? 'Something changed or could not be verified. Re-check the configured integrations for details.'
+            : 'Setup is already complete. Re-check only when troubleshooting or after external changes.',
+          'caption',
+        ),
+      );
+      verify.append(
+        verifyText,
+        actionButton(
+          current?.stack?.state === 'attention' ? 'Check now' : 'Re-check health',
+          () => readOnlyOperation('verify'),
+          'secondary',
+        ),
+      );
+      root.append(verify);
+    }
+
     const updates = node('article', undefined, 'maintenance-row');
     const updateText = node('div');
-    updateText.append(node('strong', 'Check optimizer updates'), node('p', 'Read-only version check. Nothing is upgraded automatically.', 'caption'));
-    updates.append(updateText, actionButton('Check updates', () => readOnlyOperation('updates'), 'secondary'));
-    root.append(verify, updates);
+    const available = (current?.stack?.components || []).filter(component => component.update === 'available');
+    updateText.append(
+      node('strong', 'Optimizer updates'),
+      node(
+        'p',
+        available.length
+          ? available.map(component => (TOOL_INFO[component.providerId]?.name || component.providerId) + ' has an update ready.').join(' ')
+          : 'Check installed optimizer versions. If a reviewed update is available, you can install it from the same dialog.',
+        'caption',
+      ),
+    );
+    updates.append(
+      updateText,
+      actionButton(available.length ? 'Review updates' : 'Check for updates', () => readOnlyOperation('updates'), available.length ? '' : 'secondary'),
+    );
+    root.append(updates);
+
     if (activityState?.canUndo) {
       const undo = node('article', undefined, 'maintenance-row');
       const undoText = node('div');
-      undoText.append(node('strong', 'Undo last dashboard change'), node('p', 'Review and restore the exact last transaction from this app session.', 'caption'));
+      undoText.append(
+        node('strong', 'Undo last change'),
+        node('p', 'Review and restore the exact last configuration transaction from this app session.', 'caption'),
+      );
       undo.append(undoText, actionButton('Review undo', undoLastChange, 'secondary'));
       root.append(undo);
     }
@@ -36798,42 +37072,57 @@ ${GUIDE_CANDIDATE_READINESS_JS}`;
 <header class="app-header"><div class="header-inner">
 <div class="brand"><span class="brand-mark" aria-hidden="true">TH</span><span>Token Harness<small>Use less. Know what changed.</small></span></div>
 <nav class="view-tabs" role="tablist" aria-label="Token Harness">
-<button id="tab-dashboard" role="tab" aria-selected="true" aria-controls="view-dashboard" data-view="dashboard" type="button">Dashboard</button>
-<button id="tab-setup" role="tab" aria-selected="false" aria-controls="view-setup" data-view="setup" tabindex="-1" type="button">Setup</button>
+<button id="tab-dashboard" role="tab" aria-selected="true" aria-controls="view-dashboard" data-view="dashboard" type="button">Overview</button>
 <button id="tab-results" role="tab" aria-selected="false" aria-controls="view-results" data-view="results" tabindex="-1" type="button">Results</button>
 </nav>
 <div class="header-tools"><label><span class="sr-only">Appearance</span><select id="theme" aria-label="Appearance"><option value="system">System</option><option value="light">Light</option><option value="dark">Dark</option></select></label><button id="refresh" class="secondary" type="button">Refresh</button></div>
 </div></header>
 <main id="main">
-<div class="page-heading"><div><h1 id="view-title">Dashboard</h1><p id="view-description">Your coding setup, what is active, and what to do next.</p></div></div>
+<div class="page-heading"><div><h1 id="view-title">Overview</h1><p id="view-description">Your coding agents, optimizer setup, health and measured results in one place.</p></div></div>
 <div class="status-line"><span class="read-status" role="status"><span id="reading-spinner" class="spinner" aria-hidden="true"></span><span id="updated">Checking your setup\u2026</span></span><span id="live-status" class="caption">Nothing changes without your approval</span></div>
 <div id="stale-state" class="stale-state" hidden></div><div id="error" class="error" role="alert" hidden></div><div id="notices"></div>
 
 <section id="view-dashboard" role="tabpanel" aria-labelledby="tab-dashboard" tabindex="0">
-<section id="dashboard-status" class="dashboard-status"><div><span class="eyebrow">SETUP STATUS</span><h2>Checking your machine\u2026</h2><p>Finding Claude Code, Codex, managed optimizers and measured results.</p></div><span class="pill">Checking</span></section>
-<div class="section-title"><div><h2>What Token Harness is doing for you</h2><p>Only measured results or explicitly unknown values are shown.</p></div></div>
+<section id="dashboard-status" class="dashboard-status"><div><span class="eyebrow">CURRENT STATUS</span><h2>Checking your machine\u2026</h2><p>Finding Claude Code, Codex, recommended optimizers and measured results.</p></div><span class="pill">Checking</span></section>
+
+<div class="section-title"><div><h2>Measured impact</h2><p>A summary of what Token Harness can actually prove. Open Results for the detailed evidence.</p></div></div>
 <div id="dashboard-metrics" class="impact-grid" aria-label="Efficiency summary">
 <article class="metric-card"><span class="metric-label">Measured output</span><strong class="metric-value">Checking\u2026</strong><p class="metric-help">Recorded optimizer output.</p></article>
 <article class="metric-card"><span class="metric-label">5h / 7d allowance</span><strong class="metric-value">Checking\u2026</strong><p class="metric-help">Shown only when measured.</p></article>
 <article class="metric-card"><span class="metric-label">API cost</span><strong class="metric-value">Checking\u2026</strong><p class="metric-help">Shown only from billing evidence.</p></article>
 <article class="metric-card"><span class="metric-label">Quality</span><strong class="metric-value">Checking\u2026</strong><p class="metric-help">Measured separately from savings.</p></article>
 </div>
-<div class="dashboard-columns">
-<section class="panel"><h2>Active today</h2><p class="caption">Coding agents and managed optimization that Token Harness can actually see.</p><div id="dashboard-active" class="status-list"><p class="empty">Checking\u2026</p></div></section>
-<section class="panel"><h2>Getting started</h2><p class="caption">The shortest path from installation to useful measured results.</p><div id="dashboard-checklist" class="checklist"><p class="empty">Checking\u2026</p></div></section>
-</div>
-</section>
 
-<section id="view-setup" role="tabpanel" aria-labelledby="tab-setup" tabindex="0" hidden>
-<div class="setup-intro"><div><h2>Set up Token Harness in order</h2><p>Start with your coding agent and the RTK + HarnessTrim production baseline. Optional managed integrations are reviewed separately; unmanaged experiments stay outside the active stack.</p></div><span class="pill good">Safe previews before changes</span></div>
+<section class="setup-step" id="coding-agents">
+<div class="section-title"><div><h2>Coding agents</h2><p>These are the harnesses Token Harness can configure today. A detected agent is not considered fully set up until the recommended RTK + HarnessTrim baseline is connected to it.</p></div></div>
+<div id="setup-agents" class="tool-grid"><article class="tool-card"><h3>Checking agents\u2026</h3></article></div>
+<details class="disclosure advanced-disclosure">
+<summary>Agent details and optional reasoning settings</summary>
+<p class="caption">Allowance, connected-tool observations and reasoning preferences are advanced information. They are not required to finish optimizer setup.</p>
+<div id="agent-capabilities" class="tool-grid"><article class="tool-card"><h3>Checking agent details\u2026</h3></article></div>
+<div class="subsection-heading"><h3>Optional reasoning settings</h3><p class="caption">Persistent agent preferences are separate from optimizer setup.</p></div>
+<div id="agent-tuning" class="tool-grid"><article class="tool-card"><h3>Checking reasoning controls\u2026</h3></article></div>
+</details>
+</section>
 
-<section class="setup-step"><div class="step-heading"><span class="step-number">1</span><div><h2>Coding agents</h2><p>Token Harness currently supports guided setup for Claude Code and Codex. The app must be started from an environment where the agent is detectable.</p></div></div><div id="setup-agents" class="tool-grid"><article class="tool-card"><h3>Checking agents\u2026</h3></article></div><div class="subsection-heading"><h3>Agent details</h3><p class="caption">In-session guidance, plan allowance and useful connected-tool checks.</p></div><div id="agent-capabilities" class="tool-grid"><article class="tool-card"><h3>Checking agent details\u2026</h3></article></div></section>
+<section class="setup-step" id="optimizers">
+<div class="section-title"><div><h2>Optimizers</h2><p>Recommended and optional tools are kept separate. Every button sits next to the tool or agent it affects, and every configuration change is previewed before approval.</p></div></div>
+<div class="subsection-heading"><h3>Recommended baseline</h3><p class="caption">RTK + HarnessTrim are configured together from the coding-agent card above, so first-run setup stays one action per agent.</p></div>
+<div id="managed-tools" class="tool-grid"><article class="tool-card"><h3>Checking recommended optimizers\u2026</h3></article></div>
 
-<section class="setup-step"><div class="step-heading"><span class="step-number">2</span><div><h2>Managed optimizers</h2><p>RTK and HarnessTrim are the production baseline. mcptoon, GitNexus and Headroom also expose narrow optional managed integrations on exact reviewed rows. Every setup is reviewed before anything is written.</p></div></div><div id="managed-tools" class="tool-grid"><article class="tool-card"><h3>Checking managed optimizers\u2026</h3></article></div><div id="managed-setup-actions" class="managed-action-box"><p>Checking available agents\u2026</p></div><details class="disclosure"><summary>Remove managed changes</summary><p class="caption">Removal is reviewed first. Token Harness removes only configuration changes it owns; an installed provider CLI may remain.</p><div id="managed-removal-actions" class="maintenance-list"><p class="caption">Checking managed ownership\u2026</p></div></details></section>
+<details class="disclosure optional-disclosure">
+<summary>Optional optimizers</summary>
+<p class="caption">Add these only when you want their specific capability. They are not required to complete first-run setup and are never counted as savings merely because they are configured.</p>
+<div id="optional-tools" class="tool-grid"><article class="tool-card"><h3>Checking optional optimizers\u2026</h3></article></div>
+</details>
 
-<section class="setup-step"><div class="step-heading"><span class="step-number">3</span><div><h2>Optional agent tuning</h2><p>Reasoning preferences are separate from optimizer installation. Change them only when you intentionally want a persistent agent preference.</p></div></div><div id="agent-tuning" class="tool-grid"><article class="tool-card"><h3>Checking reasoning controls\u2026</h3></article></div></section>
+<details class="disclosure"><summary>Remove Token Harness-managed configuration</summary><p class="caption">Removal is reviewed first. Token Harness removes only configuration it owns; installed provider software and user-owned configuration may remain.</p><div id="managed-removal-actions" class="maintenance-list"><p class="caption">Checking managed ownership\u2026</p></div></details>
+</section>
 
-<section class="setup-step"><div class="step-heading"><span class="step-number">4</span><div><h2>Checks and maintenance</h2><p>Read-only verification and update checks live here. Undo is shown only when this dashboard has an exact transaction it can safely restore.</p></div></div><div id="maintenance-actions" class="maintenance-list"><p class="empty">Checking\u2026</p></div></section>
+<section class="setup-step" id="maintenance">
+<div class="section-title"><div><h2>Health and updates</h2><p>These are maintenance actions, not onboarding steps. Setup already performs its own safety checks.</p></div></div>
+<div id="maintenance-actions" class="maintenance-list"><p class="empty">Checking\u2026</p></div>
+</section>
 </section>
 
 <section id="view-results" role="tabpanel" aria-labelledby="tab-results" tabindex="0" hidden>
@@ -36844,7 +37133,7 @@ ${GUIDE_CANDIDATE_READINESS_JS}`;
 <div class="section-title"><div><h2>Recent activity</h2><p>Checks and changes from this local app session.</p></div></div><section class="panel"><div id="activity"><p class="empty">No activity yet.</p></div></section>
 </section>
 
-<footer><span>Local data. No account required.</span><span>Full setup refresh runs only when you choose Refresh.</span></footer>
+<footer><span>Local data. No account required.</span><span>Full overview refresh runs only when you choose Refresh.</span></footer>
 </main>
 <dialog id="modal" aria-labelledby="modal-title"><div class="dialog-body"><div class="dialog-heading"><h2 id="modal-title">Review</h2></div><div id="modal-content"></div><div id="modal-error" class="error" role="alert" hidden></div><div id="modal-actions" class="dialog-actions"></div></div></dialog>
 <script src="/guide.js" defer></script></body></html>`;