jules-orchestrator-kit 0.35.0 → 0.35.2

package/README.md

@@ -122,7 +122,7 @@ Autonomous coding agents can write software at 100× human speed—but unconstra
 
 * **🚀 Zero-Test Bootstrapping (`agentctl bootstrap`):** Synthesizes deterministic syntax-check and smoke-test verification oracles for untested legacy repositories so agents always operate against a falsifiable feedback loop.
 
-* **📈 Proven Scale & Reliability:** Empirically tested with **526 unit tests across 79 suites passing in < 10.0s**. An adversarial red-team suite (`test/adversarial-claims.test.mjs`) continuously attempts to falsify the safety guarantees documented above — including cross-platform probes for the case-insensitive filesystems on macOS and Windows — and a documentation-sync gate (`scripts/doc-sync-check.mjs`) blocks any release whose docs have drifted from the code.
+* **📈 Proven Scale & Reliability:** Empirically tested with **529 unit tests across 79 suites passing in < 10.0s**. An adversarial red-team suite (`test/adversarial-claims.test.mjs`) continuously attempts to falsify the safety guarantees documented above — including cross-platform probes for the case-insensitive filesystems on macOS and Windows — and a documentation-sync gate (`scripts/doc-sync-check.mjs`) blocks any release whose docs have drifted from the code.
 
 <br/>
 
@@ -389,7 +389,7 @@ Native stdio server exposing task dispatch, gate verification, and risk auditing
 | :--- | :--- | :--- | :--- |
 | `init` | `agentctl init [--interactive] [--tier pro]` | Interactive onboarding wizard & stack oracle inspector generating `.agent/config.yml`. | `0` (Created) |
 | `task create` | `agentctl task create [--title <t>] [--prompt <p>] [--template <id>] [--role <name>] [--tier fast\|complex] [--depends-on <id,...>]` | Interactively authors & scopes falsifiable task envelopes with secret scrubbing, preflight gate checks, specialist role resolution, DAG dependency wiring, and an optional Cost Router tier override. | `0` (Queued), `1` (Unfalsifiable / Secret leak) |
-| `task template` | `agentctl task template [<id>] [--list] [--json]` | Lists and synthesizes specialized web task envelopes (`web-cwv`, `web-wcag`, `web-seo`, `web-playwright`, `web-flaky-heal`). | `0` (Synthesized/Listed) |
+| `task template` | `agentctl task template [<id>] [--list] [--json]` | Lists and synthesizes specialized web task envelopes (`web-cwv`, `web-wcag`, `web-seo`, `web-playwright`, `web-flaky-heal`, `web-i18n`). | `0` (Synthesized/Listed) |
 | `task optimize` | `agentctl task optimize "<prompt>" [--fix] [--web] [--json]` | Linter & optimizer injecting Google Labs 3-phase exploration budgets, critic steering, and web oracles. | `0` (Scored/Fixed) |
 | `test-gen` | `agentctl test-gen --title <t> --spec <s> [--run]` | Scaffolds falsifiable unit tests, verifies **RED** failure state, and locks test in `scope.deny`. | `0` (Scaffolded/Red) |
 | `rollback` | `agentctl rollback [sessionId \| --latest]` | Restores exact commit, uncommitted files, and cleans orphan task worktrees from pre-flight checkpoints. | `0` (Restored), `1` (Error) |

package/bin/agentctl.mjs

@@ -889,6 +889,12 @@ async function main() {
           console.log(`   Session ID   : ${sessionId}`);
           console.log(`   Digest Count : ${res.digestCount || 1}`);
           console.log(`   (Use 'agentctl escalate --flush' to deliver immediately)\n`);
+        } else if (res.dryRun) {
+          // Nothing left the machine, so do not claim it did.
+          console.log(`\n🧪 Dry run — this incident would be sent immediately:`);
+          console.log(`   Session ID : ${sessionId}`);
+          console.log(`   Reason     : ${values.reason}`);
+          console.log(`   (No request sent, and your hourly interruption budget is untouched.)\n`);
         } else if (res.dispatched) {
           console.log(`\n🚨 Incident Escalation Dispatched!`);
           console.log(`   Session ID : ${sessionId}`);

package/index.mjs

@@ -114,7 +114,11 @@ export {
   getEscalationDigestStatus,
   clearEscalationDigest,
   bufferEscalationIncident,
+  loadEscalationDigest,
+  recordInterruption,
+  countRecentInterruptions,
   DEFAULT_CRITICAL_REASONS,
+  DIGEST_BATCH_LIMIT,
 } from "./src/webhook.mjs";
 
 // PR Review Evidence Bundler & Dev Server Probe

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "jules-orchestrator-kit",
-  "version": "0.35.0",
+  "version": "0.35.2",
   "description": "Orchestration kit for running Google Jules autonomous agents.",
   "repository": {
     "type": "git",

package/src/config.mjs

@@ -349,6 +349,18 @@ export const VENDOR_TIERS = ["free", "pro", "ultra"];
 /** The tier used when a config names one that does not exist. */
 export const FALLBACK_TIER = "ultra";
 
+/**
+ * Escalation reasons that bypass the Silence Governor and alert immediately.
+ *
+ * Kept here rather than in webhook.mjs because `loadConfig` needs it as the
+ * default for `notifications.critical_reasons`, and webhook.mjs already imports
+ * from this module — the reverse direction would be a cycle. Two hand-copied
+ * lists were how v0.35.0 ended up with a governor that governed nothing.
+ *
+ * See webhook.mjs for why the list is this short.
+ */
+export const DEFAULT_CRITICAL_REASONS = ["R3_GATE_VIOLATION", "SECRET_LEAK_DETECTED", "CRITICAL_FAILURE"];
+
 /**
  * Loads and validates configuration from .agent/config.yml or .agent/jules.yml.
  */
@@ -440,7 +452,7 @@ export function loadConfig(root = resolveRoot(), explicitPath = null) {
         : 3,
       criticalReasons: Array.isArray(parsed.notifications?.critical_reasons)
         ? parsed.notifications.critical_reasons
-        : ["R3_GATE_VIOLATION", "AWAITING_USER_FEEDBACK", "OODA_REPAIR_EXHAUSTED", "SECRET_LEAK_DETECTED", "CRITICAL_FAILURE"],
+        : [...DEFAULT_CRITICAL_REASONS],
       slackWebhookUrl: parsed.notifications?.slack_webhook_url || parsed.notifications?.slackWebhookUrl || "",
       discordWebhookUrl: parsed.notifications?.discord_webhook_url || parsed.notifications?.discordWebhookUrl || "",
     },

package/src/web-templates.mjs

@@ -193,6 +193,46 @@ export const WEB_TEMPLATES = {
 5. **Stability Verification**:
    - Test suite must pass cleanly across ${reps} consecutive runs without a single failure.`;
     }
+  },
+
+  "web-i18n": {
+    id: "web-i18n",
+    name: "Internationalization (i18n), Locale Routing & Hreflang Integrity",
+    description: "Verify multi-language routing, symmetric hreflang alternate links, html lang attributes, and translation fallback safety.",
+    defaultVerifyCmd: "npm test",
+    category: "Internationalization & SEO",
+    criticFocus: [
+      "Confirm that hreflang alternate tags are fully bidirectional and symmetric across all locale variants.",
+      "Check that the <html> tag dynamically renders the correct ISO 639-1 / BCP 47 lang attribute for the active route.",
+      "Verify that localized URL slugs and route parameters handle non-ASCII unicode characters cleanly without encoding errors.",
+      "Ensure missing translation keys resolve to configured fallback locales rather than displaying raw placeholder strings or throwing runtime exceptions."
+    ],
+    defaultParams: {
+      targetLocales: "en, sv, de, fr",
+      defaultLocale: "en",
+      targetRoutes: "all public routes"
+    },
+    generatePrompt: (params = {}) => {
+      const locales = params.targetLocales || "en, sv, de, fr";
+      const defaultLoc = params.defaultLocale || "en";
+      const routes = params.targetRoutes || "all public routes";
+      const customGoal = params.goal ? `\n- **Target Focus**: ${params.goal}` : "";
+
+      return `Implement and verify comprehensive Internationalization (i18n) and locale routing integrity for ${routes}.${customGoal}
+
+### i18n & Hreflang Acceptance Criteria:
+1. **Symmetric Hreflang Alternate Links**:
+   - Every page across target locales (${locales}) must render bidirectional \`<link rel="alternate" hreflang="..." href="...">\` tags pointing to all counterparts, including self and \`hreflang="x-default"\` (pointing to ${defaultLoc}).
+2. **HTML Root & Metadata Localization**:
+   - The root \`<html lang="...">\` attribute must accurately match the active page locale.
+   - Page \`<title>\`, \`<meta name="description">\`, and OpenGraph (\`og:locale\` / \`og:locale:alternate\`) tags must be fully localized.
+3. **Locale Routing & Missing Translation Fallback**:
+   - Clean URL routing for all supported locales with zero redirect loops or 404 dead-ends.
+   - Missing translation keys must gracefully fall back to default locale (\`${defaultLoc}\`) without runtime panics or raw string delimiters.
+4. **Encoding & Number/Date Formats**:
+   - UTF-8 clean encoding with zero unescaped unicode artefacts.
+   - Localized formatting for dates, currencies, and numbers using standard Intl APIs.`;
+    }
   }
 };
 

package/src/webhook.mjs

@@ -3,16 +3,41 @@ import { createServer } from "node:http";
 import { readFileSync, writeFileSync, existsSync, unlinkSync } from "node:fs";
 import { join } from "node:path";
 import { getStateDir } from "./state.mjs";
-import { resolveRoot } from "./config.mjs";
+import { resolveRoot, DEFAULT_CRITICAL_REASONS } from "./config.mjs";
 import { redactSecrets } from "./security.mjs";
 
-export const DEFAULT_CRITICAL_REASONS = [
-  "R3_GATE_VIOLATION",
-  "AWAITING_USER_FEEDBACK",
-  "OODA_REPAIR_EXHAUSTED",
-  "SECRET_LEAK_DETECTED",
-  "CRITICAL_FAILURE",
-];
+/**
+ * Reasons that bypass the governor and page the operator the moment they occur.
+ * Defined in config.mjs (loadConfig needs the same list); re-exported here
+ * because this is the module that acts on it.
+ *
+ * The list is deliberately short. Anything named here is exempt from batching,
+ * so a reason belongs on it only when a delayed alert would let damage widen:
+ * a leaked credential keeps being valid, a violated gate keeps merging. Those
+ * are safety events, and the cost of waking someone beats the cost of waiting.
+ *
+ * `AWAITING_USER_FEEDBACK` is deliberately NOT here, even though it is the most
+ * urgent-*feeling* reason. It is the one a blocked agent raises, so on a swarm
+ * of fifteen workers it is also the most frequent by a wide margin — and it is
+ * the exact case the governor exists to batch. Listing it (as v0.35.0 did) made
+ * every default-configured escalation critical and left the governor governing
+ * nothing. Same for `OODA_REPAIR_EXHAUSTED`: the task has already stopped, so
+ * nothing worsens while it sits in a digest.
+ *
+ * An operator who wants the old behaviour sets `notifications.critical_reasons`
+ * in `.agent/config.yml` — this is a default, not a policy.
+ */
+export { DEFAULT_CRITICAL_REASONS };
+
+/**
+ * How many incidents one flush may carry.
+ *
+ * Slack truncates the summary block and Discord accepts a bounded field list,
+ * so a flush of fifty would render ten and drop forty. The digest promises the
+ * opposite — that a buffered incident is never lost — so a flush sends at most
+ * this many and leaves the remainder buffered for the next one.
+ */
+export const DIGEST_BATCH_LIMIT = 10;
 
 export function getDigestFilePath(root = resolveRoot()) {
   return join(getStateDir(root), "escalation-digest.json");
@@ -144,16 +169,21 @@ export async function flushEscalationDigest(config = {}, options = {}) {
   const discordUrl = process.env.DISCORD_WEBHOOK_URL || notifications.discordWebhookUrl || config.discordWebhookUrl || "";
   const dryRun = options.dryRun || config.dryRun || false;
 
-  const count = digest.incidents.length;
-  const results = { flushed: true, count, slack: false, discord: false, dryRun };
+  // One flush carries a bounded batch; whatever does not fit stays buffered.
+  const batch = digest.incidents.slice(0, DIGEST_BATCH_LIMIT);
+  const remainder = digest.incidents.slice(DIGEST_BATCH_LIMIT);
+  const count = batch.length;
+  const results = { flushed: true, count, pending: remainder.length, slack: false, discord: false, dryRun };
 
   if (dryRun) {
-    if (!options.preserve) clearEscalationDigest(root);
+    // A dry run shows what *would* be sent. It must leave the buffer exactly as
+    // it found it — clearing here (as v0.35.0 did) discarded real incidents in
+    // exchange for a preview.
     return {
       ...results,
       payload: {
         count,
-        incidents: digest.incidents,
+        incidents: batch,
       },
     };
   }
@@ -165,7 +195,7 @@ export async function flushEscalationDigest(config = {}, options = {}) {
   // Format Slack digest
   if (slackUrl) {
     try {
-      const summaryText = digest.incidents
+      const summaryText = batch
         .map((inc) => `• *\`${inc.sessionId}\`* on \`${inc.branch}\` [${inc.reason}]: \`agentctl resume ${inc.sessionId}\``)
         .join("\n");
 
@@ -183,7 +213,12 @@ export async function flushEscalationDigest(config = {}, options = {}) {
           {
             type: "context",
             elements: [
-              { type: "mrkdwn", text: `Aggregated by Type III Silence Governor · Oldest: ${digest.createdAt || "N/A"}` },
+              {
+                type: "mrkdwn",
+                text:
+                  `Aggregated by Type III Silence Governor · Oldest: ${digest.createdAt || "N/A"}` +
+                  (remainder.length ? ` · ${remainder.length} still buffered` : ""),
+              },
             ],
           },
         ],
@@ -203,7 +238,7 @@ export async function flushEscalationDigest(config = {}, options = {}) {
   // Format Discord digest
   if (discordUrl) {
     try {
-      const fields = digest.incidents.slice(0, 10).map((inc) => ({
+      const fields = batch.map((inc) => ({
         name: `Session \`${inc.sessionId}\` [${inc.reason}]`,
         value: `Branch: \`${inc.branch}\`\n\`agentctl resume ${inc.sessionId} --response "<answer>"\``,
         inline: false,
@@ -216,7 +251,11 @@ export async function flushEscalationDigest(config = {}, options = {}) {
             title: `Escalation Digest (${count} incidents)`,
             color: 3447003,
             fields,
-            footer: { text: `Aggregated by Type III Silence Governor · Total: ${count}` },
+            footer: {
+              text:
+                `Aggregated by Type III Silence Governor · Total: ${count}` +
+                (remainder.length ? ` · ${remainder.length} still buffered` : ""),
+            },
           },
         ],
       };
@@ -232,8 +271,13 @@ export async function flushEscalationDigest(config = {}, options = {}) {
     }
   }
 
-  if (results.slack || results.discord) {
-    if (!options.preserve) {
+  // Only what actually reached a webhook is dropped. A failed delivery leaves
+  // the whole buffer intact so the next flush retries it rather than silently
+  // eating the incidents.
+  if ((results.slack || results.discord) && !options.preserve) {
+    if (remainder.length) {
+      saveEscalationDigest(root, { incidents: remainder, createdAt: digest.createdAt });
+    } else {
       clearEscalationDigest(root);
     }
   }
@@ -303,11 +347,6 @@ export async function dispatchEscalation(incident = {}, config = {}) {
     }
   }
 
-  // If immediate/critical dispatch:
-  if (!isCritical) {
-    recordInterruption(root);
-  }
-
   const slackUrl = process.env.SLACK_WEBHOOK_URL || notifications.slackWebhookUrl || config.slackWebhookUrl || "";
   const discordUrl = process.env.DISCORD_WEBHOOK_URL || notifications.discordWebhookUrl || config.discordWebhookUrl || "";
 
@@ -341,6 +380,14 @@ export async function dispatchEscalation(incident = {}, config = {}) {
     };
   }
 
+  // The budget counts interruptions, not intentions. Recording it any earlier
+  // charged the operator's hourly allowance for a `--dry-run` preview or for a
+  // repo with no webhook configured — an alert nobody ever received. Same
+  // mistake the daily task budget made before cd26d6e; same fix.
+  if (!isCritical) {
+    recordInterruption(root);
+  }
+
   // Dispatch to Slack
   if (slackUrl) {
     try {