axstack 0.54.0 → 0.55.0

package/skills/axstack/scripts/fleet-sample.js

@@ -0,0 +1,48 @@
+import { parseProc, pressure } from './fleet-proc.js';
+import { validateManifest, sampleOptions } from './fleet-sample-manifest.js';
+import { readProc, wait } from './fleet-sample-io.js';
+export { parseProc, pressure, validateManifest };
+
+// Read functions and clocks are injectable for deterministic offline verification.
+export async function sampleFleet(manifest, hosts, options, dependencies = {}) {
+  validateManifest(manifest, hosts);
+  const { intervalMs, timeoutMs } = sampleOptions(options);
+  const { clock = Date.now, monotonic = () => performance.now(), read = readProc, wait: pause = wait,
+    signal } = dependencies;
+  const telemetry = await Promise.all(manifest.hosts.map(async (host) => {
+    const controller = new AbortController();
+    const cancel = () => controller.abort(Error('cancelled'));
+    signal?.addEventListener('abort', cancel, { once: true });
+    if (signal?.aborted) cancel();
+    const timer = setTimeout(() => controller.abort(Error('timeout')), timeoutMs);
+    const windowStartedAt = clock(); const started = monotonic();
+    const source = host.transport === 'local' ? 'linux-proc-local' : 'linux-proc-ssh';
+    const provenance = { hostId: host.hostId, environmentId: host.environmentId, scope: 'whole-host', source,
+      intervalMs, timeoutMs, windowStartedAt, windowKind: 'collector-bracket' };
+    const check = () => { if (controller.signal.aborted) throw controller.signal.reason; };
+    try {
+      check();
+      const first = parseProc(await read(host, controller.signal));
+      check();
+      await pause(intervalMs, controller.signal);
+      check();
+      const last = parseProc(await read(host, controller.signal));
+      check();
+      const windowEndedAt = clock(); const durationMs = Math.floor(monotonic() - started);
+      if (durationMs >= timeoutMs) throw Error('timeout');
+      if (durationMs < intervalMs || windowEndedAt < windowStartedAt
+        || windowEndedAt - windowStartedAt < durationMs
+        || windowEndedAt - windowStartedAt > durationMs + 1000) throw Error('collector-clock');
+      return { ...provenance, status: 'ok', ...pressure(first, last), durationMs,
+        windowEndedAt, receivedAt: windowEndedAt };
+    } catch (error) {
+      const reason = controller.signal.aborted ? controller.signal.reason?.message : error.message;
+      return { ...provenance, status: ['timeout', 'cancelled', 'malformed', 'inconsistent-counters', 'collector-clock']
+        .includes(reason) ? reason : 'failed', durationMs: Math.max(0, Math.floor(monotonic() - started)), receivedAt: clock() };
+    } finally {
+      clearTimeout(timer);
+      signal?.removeEventListener('abort', cancel);
+    }
+  }));
+  return { schemaVersion: 1, telemetry };
+}

package/skills/axstack/scripts/resolve-models.js

@@ -3,6 +3,7 @@ import { readFileSync } from 'node:fs';
 const bindings = { codex: 'codex', claude: 'claudeAgent', grok: 'grok' };
 const effortIds = { codex: 'reasoningEffort', claude: 'effort', grok: 'reasoningEffort' };
 const nonempty = (value) => typeof value === 'string' && value.trim().length > 0;
+const optionValues = (option) => option.values ?? option.options?.map((value) => value.id);
 
 function argumentsFrom(args) {
   const options = { excluded: [] };
@@ -37,9 +38,11 @@ function providerFrom(catalog, provider) {
   }
   const instance = matches[0];
   if (!Array.isArray(instance.models) || instance.models.some((model) =>
-    !model || !nonempty(model.id) || !Array.isArray(model.options) || model.options.some((option) =>
+    !model || !nonempty(model.id) || !Array.isArray(model.options ?? []) || (model.options ?? []).some((option) =>
       !option || !nonempty(option.id) || (option.options !== undefined &&
-        (!Array.isArray(option.options) || option.options.some((value) => !value || !nonempty(value.id))))))) {
+        (!Array.isArray(option.options) || option.options.some((value) => !value || !nonempty(value.id))))
+      || (option.values !== undefined && (!Array.isArray(option.values)
+        || option.values.some((value) => typeof value !== 'boolean' && !nonempty(value))))))) {
     throw new Error('malformed capabilities models or options');
   }
   return instance;
@@ -58,6 +61,7 @@ function modelFrom(models, { provider, model: pin, class: modelClass, excluded }
     }
     const model = models[0];
     if (!model || excluded.includes(model.id)) throw new Error('missing first listed model');
+    if (!model.id.startsWith('grok-')) throw new Error(`first listed model ${model.id} is not grok-family`);
     return model;
   }
   if (!['codex', 'claude'].includes(provider) || !/^[a-z]+$/.test(modelClass)) {
@@ -90,8 +94,8 @@ try {
   if ((options.class === 'terra' || /^gpt-\d+(?:\.\d+)*-terra$/.test(model.id)) && effort !== 'high') {
     throw new Error('Terra requires exactly high effort');
   }
-  const effortOptions = model.options.filter((option) => option.id === effortIds[provider]);
-  if (effortOptions.length !== 1 || !effortOptions[0].options?.some((value) => value.id === effort)
+  const effortOptions = (model.options ?? []).filter((option) => option.id === effortIds[provider]);
+  if (effortOptions.length !== 1 || !optionValues(effortOptions[0])?.includes(effort)
     || (provider === 'grok' && effort === 'max')) {
     throw new Error(`unsupported effort ${effort} for ${provider}/${model.id}`);
   }

package/skills/axstack-review/references/review-candidate.md

@@ -137,7 +137,7 @@ This section applies to peer and authored PR modes.
    For an `axstack-verify create` PR, the reviewer drives the candidate's recipe because the base has none.
    The author never performs this review drive.
    The reviewer drives each affected feature and records that drive as the `Final-head check`.
-   Delegate browser drives to `axstack-ui-verifier` with the feature file path.
+   Delegate browser drives to `axstack-ui-verifier` with the feature file path; device drives follow [UI verification](../../axstack/references/ui-verification.md#android-device-and-webpwa-dual-run)'s dual-run trigger.
    Running the applicable surface counts as spin-up; a CLI run qualifies.
    Use loopback and isolated instances under [axstack-verify](../../axstack-verify/SKILL.md#isolate-every-attempt).
 

package/skills/axstack-verify/SKILL.md

@@ -49,6 +49,13 @@ Name the app, surface and invocation conditions in the description.
 - Cleanup: stop only processes started by this run and remove owned scratch.
 - Helpers: explain each helper's purpose and working directory.
 
+For a device surface, require a Device section in generated `verify-<app>` recipes.
+The Device section must use device tools to install and launch the pinned build,
+drive features, capture evidence and shutdown owned devices.
+The Device section must follow [UI verification](../axstack/references/ui-verification.md#android-device-and-webpwa-dual-run)
+for dual-run independence and reconciliation, fleet routing and desktop safety.
+Record unavailable device surfaces as unverified.
+
 Write `.agents/skills/verify-<app>/features/README.md` as an index of the top 3-5 user features.
 Write per-feature files in `.agents/skills/verify-<app>/features/` and link them from the index.
 Give each feature file its entry points, prerequisites, drive steps and observable proof.
@@ -79,10 +86,12 @@ Report an unavailable candidate run as unverified.
 
 Execute launch, doctor, drive one mapped feature, capture evidence, clean up and confirm evidence survives.
 Run doctor before each drive.
-The author drives only non-browser surfaces.
-Delegate browser drives through
-[UI verification](../axstack/references/ui-verification.md) to `axstack-ui-verifier`
+The author drives only non-browser, non-device surfaces.
+Delegate a recipe's device checks and explicitly requested app/PWA verification
+to the UI verifier roles for a dual run through
+[UI verification](../axstack/references/ui-verification.md#android-device-and-webpwa-dual-run)
 with the feature file path in the brief.
+Delegate other recipe browser drives to one `axstack-ui-verifier` pass.
 Give the verifier the pinned candidate, isolated instance and private evidence path.
 Clean up after every failed attempt.
 Retry once after a drift fix.
@@ -124,10 +133,12 @@ Classify each gap as doc drift, harness gap or product regression:
 Drive every mapped feature live even when source looks clean.
 Use the skill's Launch recipe for an isolated instance.
 Run doctor before each drive.
-The author drives only non-browser surfaces.
-Delegate browser drives through
-[UI verification](../axstack/references/ui-verification.md) to `axstack-ui-verifier`
+The author drives only non-browser, non-device surfaces.
+Delegate a recipe's device checks and explicitly requested app/PWA verification
+to the UI verifier roles for a dual run through
+[UI verification](../axstack/references/ui-verification.md#android-device-and-webpwa-dual-run)
 with the feature file path in the brief.
+Delegate other recipe browser drives to one `axstack-ui-verifier` pass.
 Capture actions, resulting states and side effects under its proof standards.
 Report unreachable prerequisites with attempted routes separately from passes.
 Fix a missing prerequisite in the map as doc drift.

package/src/roles.js

@@ -92,8 +92,8 @@ export function assessRoleReadiness(roles, preset) {
   const isIntentionalAbsence = (role) => role.model === null && (
     (preset === 'mixed' && role.id === 'axstack-research-x' && role.provider === 'grok') ||
     (preset === 'mixed' && role.id === 'axstack-arena-candidate-grok' && role.provider === 'grok') ||
-    (preset === 'codex-only' && ['axstack-advisor-opus', 'axstack-escalation-fable', 'axstack-arena-judge-opus', 'axstack-research-x', 'axstack-arena-candidate-grok', 'axstack-research-web-sol'].includes(role.id)) ||
-    (preset === 'claude-only' && ['axstack-advisor-astra', 'axstack-arena-judge-astra', 'axstack-research-x', 'axstack-arena-candidate-grok', 'axstack-auditor-sol', 'axstack-research-code-sol', 'axstack-explore-execution-sol', 'axstack-explore-codebase-luna', 'axstack-research-web-sol'].includes(role.id))
+    (preset === 'codex-only' && ['axstack-advisor-opus', 'axstack-escalation-fable', 'axstack-arena-judge-opus', 'axstack-research-x', 'axstack-arena-candidate-grok', 'axstack-research-web-sol', 'axstack-ui-verifier-sonnet'].includes(role.id)) ||
+    (preset === 'claude-only' && ['axstack-advisor-astra', 'axstack-arena-judge-astra', 'axstack-research-x', 'axstack-arena-candidate-grok', 'axstack-auditor-sol', 'axstack-research-code-sol', 'axstack-explore-execution-sol', 'axstack-explore-codebase-luna', 'axstack-research-web-sol', 'axstack-ui-verifier-sonnet'].includes(role.id))
   );
   for (const role of roles) {
     if (!bounds.has(role.provider)) {