zenox 1.1.3 → 1.2.1

package/README.md

@@ -10,7 +10,6 @@
 
 <p align="center">
   <a href="https://www.npmjs.com/package/zenox"><img src="https://img.shields.io/npm/v/zenox.svg?style=flat-square" alt="npm version" /></a>
-  <!-- <a href="https://www.npmjs.com/package/zenox"><img src="https://img.shields.io/npm/dm/zenox.svg?style=flat-square" alt="npm downloads" /></a> -->
   <a href="https://github.com/CYBERBOYAYUSH/zenox/blob/main/LICENSE"><img src="https://img.shields.io/badge/license-MIT-blue.svg?style=flat-square" alt="license" /></a>
 </p>
 
@@ -41,7 +40,7 @@ That's it. Restart OpenCode and the agents are ready.
 
 | Agent | What it does | Default Model |
 |-------|-------------|---------------|
-| **Explorer** | Codebase search, file discovery, pattern matching | `claude-haiku-4-5` |
+| **Explorer** | Codebase grep, file discovery, pattern matching | `claude-haiku-4-5` |
 | **Librarian** | Library research, docs lookup, GitHub examples | `claude-sonnet-4-5` |
 | **Oracle** | Architecture decisions, debugging strategy, code review | `gpt-5.2` |
 | **UI Planner** | Frontend design, CSS, animations, visual polish | `gemini-3-pro-high` |
@@ -64,6 +63,36 @@ You: "Make this dashboard look better"
 → UI Planner redesigns with proper aesthetics
 ```
 
+## Keyword Triggers
+
+Include these magic words in your prompt to unlock special modes:
+
+| Keyword | What it does |
+|---------|--------------|
+| `ultrawork` or `ulw` | Maximum multi-agent coordination — fires parallel background agents, sets max precision |
+| `deep research` | Comprehensive exploration — fires 3-4 background agents (explorer + librarian) |
+| `explore codebase` | Codebase mapping — multiple explorers search in parallel |
+
+### Examples
+
+```
+You: "ultrawork - add authentication to this app"
+→ ⚡ Ultrawork Mode activated
+→ Fires explorer + librarian in parallel
+→ Maximum precision engaged
+
+You: "deep research how this project handles errors"
+→ 🔬 Deep Research Mode activated
+→ Fires multiple explorers + librarians
+→ Waits for comprehensive results before proceeding
+
+You: "explore codebase for payment logic"
+→ 🔍 Explore Mode activated
+→ Multiple explorers search patterns, implementations, tests
+```
+
+You'll see a toast notification when these modes activate.
+
 ## Background Tasks
 
 Need comprehensive research? Fire multiple agents in parallel:
@@ -76,6 +105,15 @@ background_task(agent="librarian", description="JWT best practices", prompt="...
 // You're notified when all tasks complete
 ```
 
+### Toast Notifications
+
+Zenox shows toast notifications for background task events:
+
+- ⚡ **Task Launched** — Shows task description and agent
+- ✅ **Task Completed** — Shows duration and remaining count
+- 🎉 **All Complete** — Shows summary of all finished tasks
+- ❌ **Task Failed** — Shows error message
+
 ## Configuration
 
 ### Custom Models
@@ -133,7 +171,7 @@ Zenox checks for updates on startup. When a new version drops:
 2. Bun cache is invalidated
 3. Restart to get the update
 
-Pin a version to disable: `"zenox@1.0.3"` in your plugins array.
+Pin a version to disable: `"zenox@1.2.0"` in your plugins array.
 
 ## Credits
 

package/dist/background/manager.d.ts

@@ -9,9 +9,12 @@
  */
 import type { OpencodeClient } from "@opencode-ai/sdk";
 import type { BackgroundTask, LaunchInput, CompletionNotification } from "./types";
+import type { TaskToastManager } from "../features/task-toast";
 export declare class BackgroundManager {
     private tasks;
     private mainSessionID;
+    private toastManager;
+    setToastManager(manager: TaskToastManager): void;
     setMainSession(sessionID: string): void;
     getMainSession(): string | undefined;
     private generateTaskId;

package/dist/features/task-toast/index.d.ts

@@ -0,0 +1 @@
+export { TaskToastManager } from "./manager";

package/dist/features/task-toast/manager.d.ts

@@ -0,0 +1,38 @@
+/**
+ * Task Toast Manager
+ *
+ * Manages toast notifications for background tasks:
+ * - Shows toast when task is launched
+ * - Shows toast when task completes
+ * - Shows summary toast when all tasks complete
+ */
+import type { OpencodeClient } from "@opencode-ai/sdk";
+interface TrackedTask {
+    id: string;
+    description: string;
+    agent: string;
+    startedAt: Date;
+    status: "running" | "completed" | "failed";
+}
+export declare class TaskToastManager {
+    private tasks;
+    private client;
+    constructor(client: OpencodeClient);
+    showLaunchToast(task: {
+        id: string;
+        description: string;
+        agent: string;
+    }): Promise<void>;
+    showCompletionToast(taskId: string): Promise<void>;
+    showFailureToast(taskId: string, error?: string): Promise<void>;
+    private showAllCompleteToast;
+    private getRunningCount;
+    private clearCompleted;
+    trackTask(task: {
+        id: string;
+        description: string;
+        agent: string;
+    }): void;
+    getTask(taskId: string): TrackedTask | undefined;
+}
+export {};

package/dist/hooks/index.d.ts

@@ -1 +1,3 @@
 export { createAutoUpdateHook } from "./auto-update";
+export { createKeywordDetectorHook } from "./keyword-detector";
+export type { KeywordType, KeywordConfig } from "./keyword-detector";

package/dist/hooks/keyword-detector/contexts.d.ts

@@ -0,0 +1,18 @@
+/**
+ * Context strings injected when keywords are detected.
+ * These provide mode-specific guidance to the agent.
+ */
+export declare const ULTRAWORK_CONTEXT = "<ultrawork-mode>\nULTRAWORK MODE ACTIVE - Maximum multi-agent coordination engaged.\n\n**Mandatory behaviors**:\n1. Fire explore + librarian in PARALLEL BACKGROUND for ANY research task\n2. Use pre-delegation reasoning before EVERY delegation\n3. Continue productive work while background agents run (never idle)\n4. Consult oracle for architectural decisions before implementing\n5. Verify ALL delegated work before marking tasks complete\n\n**Pre-delegation reasoning** (do this before every Task/background_task):\n- What type of work? (search \u2192 background, decision \u2192 sync, visual \u2192 sync)\n- Can I continue working while this runs? If yes \u2192 background_task\n- Do I need this answer RIGHT NOW to proceed? If no \u2192 background_task\n\n**Parallel research pattern**:\n```\n// Fire research immediately\nbackground_task(agent=\"explorer\", description=\"Find patterns\", prompt=\"...\")\nbackground_task(agent=\"librarian\", description=\"Find best practices\", prompt=\"...\")\n\n// Continue working while they run:\n// - Plan implementation approach\n// - Read files you know about\n// - Identify edge cases\n\n// When notified \u2192 retrieve and synthesize\nbackground_output(task_id=\"bg_xxx\")\n```\n</ultrawork-mode>";
+export declare const DEEP_RESEARCH_CONTEXT = "<deep-research-mode>\nDEEP RESEARCH MODE - Comprehensive exploration requested.\n\n**Required actions**:\n1. Fire at least 2-3 parallel background agents before proceeding\n2. Search BOTH internal (explorer) AND external (librarian) sources\n3. Explore multiple angles: implementations, tests, patterns, docs\n4. DO NOT proceed until you have comprehensive context\n5. Synthesize all findings before making decisions\n\n**Research pattern**:\n```\n// Fire comprehensive research\nbackground_task(agent=\"explorer\", description=\"Find implementations\", prompt=\"...\")\nbackground_task(agent=\"explorer\", description=\"Find related tests\", prompt=\"...\")\nbackground_task(agent=\"librarian\", description=\"Find official docs\", prompt=\"...\")\nbackground_task(agent=\"librarian\", description=\"Find OSS examples\", prompt=\"...\")\n\n// Wait for ALL results before proceeding\n// When notified \u2192 retrieve, synthesize, then act\n```\n</deep-research-mode>";
+export declare const EXPLORE_CONTEXT = "<explore-mode>\nEXPLORE MODE - Codebase exploration active.\n\n**Required actions**:\n1. Fire multiple explorer agents in background for parallel search\n2. Search for: patterns, implementations, tests, related code\n3. Map out the relevant code landscape before modifying\n4. Look for existing conventions and follow them\n\n**Exploration pattern**:\n```\n// Fire multiple explorers in parallel\nbackground_task(agent=\"explorer\", description=\"Find main implementation\", prompt=\"...\")\nbackground_task(agent=\"explorer\", description=\"Find related tests\", prompt=\"...\")\nbackground_task(agent=\"explorer\", description=\"Find usage examples\", prompt=\"...\")\n\n// Continue analyzing what you know while they search\n```\n</explore-mode>";
+export type KeywordType = "ultrawork" | "deep-research" | "explore";
+export interface KeywordConfig {
+    type: KeywordType;
+    pattern: RegExp;
+    context: string;
+    toast: {
+        title: string;
+        message: string;
+    };
+}
+export declare const KEYWORD_CONFIGS: KeywordConfig[];

package/dist/hooks/keyword-detector/index.d.ts

@@ -0,0 +1,28 @@
+/**
+ * Keyword Detector Hook
+ *
+ * Detects special keywords in user messages and:
+ * 1. Injects mode-specific context into the message
+ * 2. Sets message variant to "max" for higher precision (ultrawork only)
+ * 3. Shows toast notification for user feedback
+ *
+ * Uses the "chat.message" hook which fires when user submits a message.
+ */
+import type { PluginInput } from "@opencode-ai/plugin";
+interface MessagePart {
+    type: string;
+    text?: string;
+}
+interface ChatMessageOutput {
+    parts: MessagePart[];
+    message: {
+        variant?: string;
+    };
+}
+export declare function createKeywordDetectorHook(ctx: PluginInput): {
+    "chat.message": (input: {
+        sessionID: string;
+        agent?: string;
+    }, output: ChatMessageOutput) => Promise<void>;
+};
+export { KEYWORD_CONFIGS, type KeywordType, type KeywordConfig } from "./contexts";

package/dist/index.js

@@ -548,10 +548,21 @@ You have specialized subagents. Use the **Task tool** to delegate work proactive
 
 | Agent | Use For | subagent_type |
 |-------|---------|---------------|
-| **Explorer** | Codebase search, "Where is X?", file discovery, pattern matching | \`explorer\` |
-| **Librarian** | External docs, library research, OSS examples, GitHub permalinks | \`librarian\` |
-| **Oracle** | Architecture decisions, debugging strategy, trade-offs, code review | \`oracle\` |
-| **UI Planner** | Visual design, CSS, animations, UI/UX, beautiful interfaces | \`ui-planner\` |
+| **Explorer** | Codebase grep - fast pattern matching, "Where is X?" | \`explorer\` |
+| **Librarian** | External grep - docs, GitHub, OSS examples | \`librarian\` |
+| **Oracle** | Strategic advisor - architecture, debugging, decisions | \`oracle\` |
+| **UI Planner** | Designer-developer - visual design, CSS, animations | \`ui-planner\` |
+
+### Quick Rule: Background vs Synchronous
+
+| Agent | Default Execution | Why |
+|-------|-------------------|-----|
+| Explorer | \`background_task\` | It's codebase grep - fire and continue |
+| Librarian | \`background_task\` | It's external grep - fire and continue |
+| Oracle | \`Task\` (sync) | Need strategic answer before proceeding |
+| UI Planner | \`Task\` (sync) | Implements changes, needs write access |
+
+**Mental Model**: Explorer & Librarian = **grep commands**. You don't wait for grep, you fire it and continue thinking.
 
 ### When to Delegate (Fire Immediately)
 
@@ -699,6 +710,39 @@ background_output(task_id="bg_ghi789")
 - **Background** = Use when researching multiple angles independently
 
 **Both tools coexist - choose based on whether tasks are dependent or independent.**
+
+### The Parallel Research Pattern
+
+For complex tasks, fire research first, then continue working:
+
+\`\`\`
+// 1. FIRE parallel research (don't wait!)
+background_task(agent="explorer", description="Find existing patterns", prompt="...")
+background_task(agent="librarian", description="Find best practices", prompt="...")
+
+// 2. CONTINUE productive work while they run:
+//    - Plan your implementation approach
+//    - Read files you already know about
+//    - Identify edge cases and questions
+
+// 3. When notified \u2192 RETRIEVE and synthesize
+background_output(task_id="bg_xxx")
+background_output(task_id="bg_yyy")
+\`\`\`
+
+**Anti-pattern**: Firing background tasks then doing nothing. Always continue productive work!
+
+---
+
+## Keyword Triggers (Power User)
+
+Include these keywords in your prompt to unlock special modes:
+
+| Keyword | Effect |
+|---------|--------|
+| \`ultrawork\` or \`ulw\` | Maximum multi-agent coordination - aggressive parallel research |
+| \`deep research\` | Comprehensive exploration - fires 3-4 background agents |
+| \`explore codebase\` | Codebase mapping - multiple explorers in parallel |
 `;
 
 // node_modules/.pnpm/zod@3.25.76/node_modules/zod/v3/external.js
@@ -4801,6 +4845,10 @@ function createBuiltinMcps(disabledMcps = []) {
 class BackgroundManager {
   tasks = new Map;
   mainSessionID;
+  toastManager;
+  setToastManager(manager) {
+    this.toastManager = manager;
+  }
   setMainSession(sessionID) {
     this.mainSessionID = sessionID;
   }
@@ -4834,6 +4882,13 @@ class BackgroundManager {
       startedAt: new Date
     };
     this.tasks.set(task.id, task);
+    if (this.toastManager) {
+      this.toastManager.showLaunchToast({
+        id: task.id,
+        description: task.description,
+        agent: task.agent
+      }).catch(() => {});
+    }
     client.session.prompt({
       path: { id: sessionID },
       body: {
@@ -4847,6 +4902,9 @@ class BackgroundManager {
         existingTask.status = "failed";
         existingTask.error = err.message ?? "Unknown error";
         existingTask.completedAt = new Date;
+        if (this.toastManager) {
+          this.toastManager.showFailureToast(task.id, existingTask.error).catch(() => {});
+        }
       }
     });
     return task;
@@ -4919,6 +4977,9 @@ class BackgroundManager {
     }
     task.status = "completed";
     task.completedAt = new Date;
+    if (this.toastManager) {
+      this.toastManager.showCompletionToast(task.id).catch(() => {});
+    }
     return this.getCompletionStatus();
   }
   getCompletionStatus() {
@@ -17480,14 +17541,291 @@ async function showVersionToast(ctx, version2) {
     }
   }).catch(() => {});
 }
+// src/hooks/keyword-detector/contexts.ts
+var ULTRAWORK_CONTEXT = `<ultrawork-mode>
+ULTRAWORK MODE ACTIVE - Maximum multi-agent coordination engaged.
+
+**Mandatory behaviors**:
+1. Fire explore + librarian in PARALLEL BACKGROUND for ANY research task
+2. Use pre-delegation reasoning before EVERY delegation
+3. Continue productive work while background agents run (never idle)
+4. Consult oracle for architectural decisions before implementing
+5. Verify ALL delegated work before marking tasks complete
+
+**Pre-delegation reasoning** (do this before every Task/background_task):
+- What type of work? (search \u2192 background, decision \u2192 sync, visual \u2192 sync)
+- Can I continue working while this runs? If yes \u2192 background_task
+- Do I need this answer RIGHT NOW to proceed? If no \u2192 background_task
+
+**Parallel research pattern**:
+\`\`\`
+// Fire research immediately
+background_task(agent="explorer", description="Find patterns", prompt="...")
+background_task(agent="librarian", description="Find best practices", prompt="...")
+
+// Continue working while they run:
+// - Plan implementation approach
+// - Read files you know about
+// - Identify edge cases
+
+// When notified \u2192 retrieve and synthesize
+background_output(task_id="bg_xxx")
+\`\`\`
+</ultrawork-mode>`;
+var DEEP_RESEARCH_CONTEXT = `<deep-research-mode>
+DEEP RESEARCH MODE - Comprehensive exploration requested.
+
+**Required actions**:
+1. Fire at least 2-3 parallel background agents before proceeding
+2. Search BOTH internal (explorer) AND external (librarian) sources
+3. Explore multiple angles: implementations, tests, patterns, docs
+4. DO NOT proceed until you have comprehensive context
+5. Synthesize all findings before making decisions
+
+**Research pattern**:
+\`\`\`
+// Fire comprehensive research
+background_task(agent="explorer", description="Find implementations", prompt="...")
+background_task(agent="explorer", description="Find related tests", prompt="...")
+background_task(agent="librarian", description="Find official docs", prompt="...")
+background_task(agent="librarian", description="Find OSS examples", prompt="...")
+
+// Wait for ALL results before proceeding
+// When notified \u2192 retrieve, synthesize, then act
+\`\`\`
+</deep-research-mode>`;
+var EXPLORE_CONTEXT = `<explore-mode>
+EXPLORE MODE - Codebase exploration active.
+
+**Required actions**:
+1. Fire multiple explorer agents in background for parallel search
+2. Search for: patterns, implementations, tests, related code
+3. Map out the relevant code landscape before modifying
+4. Look for existing conventions and follow them
+
+**Exploration pattern**:
+\`\`\`
+// Fire multiple explorers in parallel
+background_task(agent="explorer", description="Find main implementation", prompt="...")
+background_task(agent="explorer", description="Find related tests", prompt="...")
+background_task(agent="explorer", description="Find usage examples", prompt="...")
+
+// Continue analyzing what you know while they search
+\`\`\`
+</explore-mode>`;
+var KEYWORD_CONFIGS = [
+  {
+    type: "ultrawork",
+    pattern: /\b(ultrawork|ulw)\b/i,
+    context: ULTRAWORK_CONTEXT,
+    toast: {
+      title: "\u26A1 Ultrawork Mode",
+      message: "Maximum precision engaged. Multi-agent coordination active."
+    }
+  },
+  {
+    type: "deep-research",
+    pattern: /\b(deep\s*research|research\s*deep(ly)?)\b/i,
+    context: DEEP_RESEARCH_CONTEXT,
+    toast: {
+      title: "\uD83D\uDD2C Deep Research Mode",
+      message: "Comprehensive exploration enabled. Background agents will fire."
+    }
+  },
+  {
+    type: "explore",
+    pattern: /\b(explore\s*(the\s*)?(codebase|code|project))\b/i,
+    context: EXPLORE_CONTEXT,
+    toast: {
+      title: "\uD83D\uDD0D Explore Mode",
+      message: "Codebase exploration active. Multiple explorers will run."
+    }
+  }
+];
+
+// src/hooks/keyword-detector/index.ts
+var TOAST_DURATION2 = 4000;
+function extractTextFromParts(parts) {
+  return parts.filter((p) => p.type === "text" && p.text).map((p) => p.text ?? "").join(" ");
+}
+function removeCodeBlocks(text) {
+  return text.replace(/```[\s\S]*?```/g, "").replace(/`[^`]+`/g, "");
+}
+function detectKeywords(text) {
+  const cleanText = removeCodeBlocks(text);
+  const detected = [];
+  for (const config2 of KEYWORD_CONFIGS) {
+    if (config2.pattern.test(cleanText)) {
+      detected.push(config2);
+    }
+  }
+  return detected;
+}
+function createKeywordDetectorHook(ctx) {
+  const detectedSessions = new Set;
+  return {
+    "chat.message": async (input, output) => {
+      const promptText = extractTextFromParts(output.parts);
+      const detectedKeywords = detectKeywords(promptText);
+      if (detectedKeywords.length === 0)
+        return;
+      const sessionKey = `${input.sessionID}-${detectedKeywords.map((k) => k.type).join("-")}`;
+      if (detectedSessions.has(sessionKey))
+        return;
+      detectedSessions.add(sessionKey);
+      const primaryKeyword = detectedKeywords[0];
+      if (primaryKeyword.type === "ultrawork") {
+        output.message.variant = "max";
+      }
+      output.parts.push({
+        type: "text",
+        text: primaryKeyword.context
+      });
+      await ctx.client.tui.showToast({
+        body: {
+          title: primaryKeyword.toast.title,
+          message: primaryKeyword.toast.message,
+          variant: "success",
+          duration: TOAST_DURATION2
+        }
+      }).catch(() => {});
+      if (detectedSessions.size > 100) {
+        const entries = [...detectedSessions];
+        entries.slice(0, 50).forEach((key) => detectedSessions.delete(key));
+      }
+    }
+  };
+}
+// src/features/task-toast/manager.ts
+var LAUNCH_TOAST_DURATION = 3000;
+var COMPLETE_TOAST_DURATION = 4000;
+var ALL_COMPLETE_TOAST_DURATION = 5000;
+function formatDuration(startedAt) {
+  const seconds = Math.floor((Date.now() - startedAt.getTime()) / 1000);
+  if (seconds < 60)
+    return `${seconds}s`;
+  const minutes = Math.floor(seconds / 60);
+  const remainingSeconds = seconds % 60;
+  return `${minutes}m ${remainingSeconds}s`;
+}
+
+class TaskToastManager {
+  tasks = new Map;
+  client;
+  constructor(client) {
+    this.client = client;
+  }
+  async showLaunchToast(task) {
+    this.tasks.set(task.id, {
+      ...task,
+      startedAt: new Date,
+      status: "running"
+    });
+    const runningCount = this.getRunningCount();
+    const message = runningCount > 1 ? `"${task.description}"
+Agent: ${task.agent}
+Running: ${runningCount}` : `"${task.description}"
+Agent: ${task.agent}`;
+    await this.client.tui.showToast({
+      body: {
+        title: "\u26A1 Background Task Launched",
+        message,
+        variant: "info",
+        duration: LAUNCH_TOAST_DURATION
+      }
+    }).catch(() => {});
+  }
+  async showCompletionToast(taskId) {
+    const task = this.tasks.get(taskId);
+    if (!task)
+      return;
+    task.status = "completed";
+    const duration3 = formatDuration(task.startedAt);
+    const runningCount = this.getRunningCount();
+    if (runningCount === 0) {
+      await this.showAllCompleteToast();
+      return;
+    }
+    const message = `"${task.description}" (${duration3})
+Still running: ${runningCount}`;
+    await this.client.tui.showToast({
+      body: {
+        title: "\u2705 Task Completed",
+        message,
+        variant: "success",
+        duration: COMPLETE_TOAST_DURATION
+      }
+    }).catch(() => {});
+  }
+  async showFailureToast(taskId, error45) {
+    const task = this.tasks.get(taskId);
+    if (!task)
+      return;
+    task.status = "failed";
+    const duration3 = formatDuration(task.startedAt);
+    const message = error45 ? `"${task.description}" (${duration3})
+Error: ${error45}` : `"${task.description}" (${duration3})`;
+    await this.client.tui.showToast({
+      body: {
+        title: "\u274C Task Failed",
+        message,
+        variant: "error",
+        duration: COMPLETE_TOAST_DURATION
+      }
+    }).catch(() => {});
+  }
+  async showAllCompleteToast() {
+    const completedTasks = [...this.tasks.values()].filter((t) => t.status === "completed" || t.status === "failed");
+    const taskList = completedTasks.map((t) => {
+      const icon = t.status === "completed" ? "\u2713" : "\u2717";
+      const duration3 = formatDuration(t.startedAt);
+      return `${icon} ${t.description} (${duration3})`;
+    }).join(`
+`);
+    await this.client.tui.showToast({
+      body: {
+        title: "\uD83C\uDF89 All Background Tasks Complete",
+        message: `${taskList}
+
+Use background_output to retrieve.`,
+        variant: "success",
+        duration: ALL_COMPLETE_TOAST_DURATION
+      }
+    }).catch(() => {});
+    this.clearCompleted();
+  }
+  getRunningCount() {
+    return [...this.tasks.values()].filter((t) => t.status === "running").length;
+  }
+  clearCompleted() {
+    for (const [id, task] of this.tasks) {
+      if (task.status !== "running") {
+        this.tasks.delete(id);
+      }
+    }
+  }
+  trackTask(task) {
+    this.tasks.set(task.id, {
+      ...task,
+      startedAt: new Date,
+      status: "running"
+    });
+  }
+  getTask(taskId) {
+    return this.tasks.get(taskId);
+  }
+}
 // src/index.ts
 var ZenoxPlugin = async (ctx) => {
   const pluginConfig = loadPluginConfig(ctx.directory);
   const disabledAgents = new Set(pluginConfig.disabled_agents ?? []);
   const disabledMcps = pluginConfig.disabled_mcps ?? [];
+  const taskToastManager = new TaskToastManager(ctx.client);
   const backgroundManager = new BackgroundManager;
+  backgroundManager.setToastManager(taskToastManager);
   const backgroundTools = createBackgroundTools(backgroundManager, ctx.client);
   const autoUpdateHook = createAutoUpdateHook(ctx);
+  const keywordDetectorHook = createKeywordDetectorHook(ctx);
   const applyModelOverride = (agentName, baseAgent) => {
     const override = pluginConfig.agents?.[agentName];
     if (override?.model) {
@@ -17497,6 +17835,9 @@ var ZenoxPlugin = async (ctx) => {
   };
   return {
     tool: backgroundTools,
+    "chat.message": async (input, output) => {
+      await keywordDetectorHook["chat.message"]?.(input, output);
+    },
     event: async (input) => {
       const { event } = input;
       await autoUpdateHook.event(input);

package/dist/orchestration/prompt.d.ts

@@ -2,4 +2,4 @@
  * Orchestration prompt to inject into Build and Plan agents.
  * This teaches the primary agents how to delegate to specialized subagents using the Task tool.
  */
-export declare const ORCHESTRATION_PROMPT = "\n\n---\n\n## Sub-Agent Delegation\n\nYou have specialized subagents. Use the **Task tool** to delegate work proactively.\n\n### Available Agents\n\n| Agent | Use For | subagent_type |\n|-------|---------|---------------|\n| **Explorer** | Codebase search, \"Where is X?\", file discovery, pattern matching | `explorer` |\n| **Librarian** | External docs, library research, OSS examples, GitHub permalinks | `librarian` |\n| **Oracle** | Architecture decisions, debugging strategy, trade-offs, code review | `oracle` |\n| **UI Planner** | Visual design, CSS, animations, UI/UX, beautiful interfaces | `ui-planner` |\n\n### When to Delegate (Fire Immediately)\n\n| Trigger | subagent_type | Why |\n|---------|---------------|-----|\n| \"Where is X?\", \"Find Y\", locate code | `explorer` | Fast codebase search |\n| External library, \"how does X library work?\" | `librarian` | Searches docs, GitHub, OSS |\n| 2+ modules involved, cross-cutting concerns | `explorer` | Multi-file pattern discovery |\n| Architecture decision, \"should I use X or Y?\" | `oracle` | Deep reasoning advisor |\n| Visual/styling, CSS, animations, UI/UX | `ui-planner` | Designer-developer hybrid |\n| After 2+ failed fix attempts | `oracle` | Debugging escalation |\n\n### How to Delegate\n\nUse the Task tool with these parameters:\n\n```\nTask(\n  subagent_type: \"explorer\" | \"librarian\" | \"oracle\" | \"ui-planner\",\n  description: \"Short 3-5 word task description\",\n  prompt: \"Detailed instructions for the agent\"\n)\n```\n\n**Example delegations:**\n\n```\n// Find code in codebase\nTask(\n  subagent_type: \"explorer\",\n  description: \"Find auth middleware\",\n  prompt: \"Find all authentication middleware implementations in this codebase. Return file paths and explain the auth flow.\"\n)\n\n// Research external library\nTask(\n  subagent_type: \"librarian\",\n  description: \"React Query caching docs\",\n  prompt: \"How does React Query handle caching? Find official documentation and real-world examples with GitHub permalinks.\"\n)\n\n// Architecture decision\nTask(\n  subagent_type: \"oracle\",\n  description: \"Redux vs Zustand analysis\",\n  prompt: \"Analyze trade-offs between Redux and Zustand for this project. Consider bundle size, learning curve, and our existing patterns.\"\n)\n\n// UI/Visual work\nTask(\n  subagent_type: \"ui-planner\",\n  description: \"Redesign dashboard cards\",\n  prompt: \"Redesign the dashboard stat cards to be more visually appealing. Use modern aesthetics, subtle animations, and ensure responsive design.\"\n)\n```\n\n### Parallel Execution\n\nTo run multiple agents in parallel, call multiple Task tools in the **same response message**:\n\n```\n// CORRECT: Multiple Task calls in ONE message = parallel execution\nTask(subagent_type: \"explorer\", description: \"Find auth code\", prompt: \"...\")\nTask(subagent_type: \"librarian\", description: \"JWT best practices\", prompt: \"...\")\n// Both run simultaneously\n\n// WRONG: One Task per message = sequential (slow)\nMessage 1: Task(...) \u2192 wait for result\nMessage 2: Task(...) \u2192 wait for result\n```\n\n### Delegation Priority\n\n1. **Explorer FIRST** \u2014 Always locate code before modifying unfamiliar areas\n2. **Librarian** \u2014 When dealing with external libraries/APIs you don't fully understand\n3. **Oracle** \u2014 For complex decisions or after 2+ failed fix attempts\n4. **UI Planner** \u2014 For ANY visual/styling work (never edit CSS/UI yourself)\n\n### Critical Rules\n\n1. **Never touch frontend visual/styling code yourself** \u2014 Always delegate to `ui-planner`\n2. **Fire librarian proactively** when unfamiliar libraries are involved\n3. **Consult oracle BEFORE major architectural decisions**, not after\n4. **Verify delegated work** before marking task complete\n\n### When to Handle Directly (Don't Delegate)\n\n- Single file edits with known location\n- Questions answerable from code already in context\n- Trivial changes requiring no specialist knowledge\n- Tasks you can complete faster than explaining to an agent\n\n---\n\n## Background Tasks (Parallel Research)\n\nFor **independent research tasks** that benefit from parallelism, use background tasks instead of sequential Task calls.\n\n### When to Use Background Tasks\n\n| Scenario | Use Background Tasks |\n|----------|---------------------|\n| User wants \"comprehensive\" / \"thorough\" / \"deep\" exploration | YES - fire 3-4 agents in parallel |\n| Need BOTH codebase search AND external docs | YES - explore + librarian in parallel |\n| Exploring multiple modules/features simultaneously | YES - separate explore for each |\n| Result of Task A needed before Task B | NO - use sequential Task |\n| Single focused lookup | NO - just use Task directly |\n\n### How Background Tasks Work\n\n1. **Fire**: Launch multiple agents with `background_task` - they run in parallel\n2. **Continue**: Keep working while background agents search\n3. **Notify**: You'll be notified when ALL background tasks complete\n4. **Retrieve**: Use `background_output` to get each result\n\n### Usage\n\n```\n// Launch parallel research (all run simultaneously)\nbackground_task(agent=\"explorer\", description=\"Find auth code\", prompt=\"Search for authentication...\")\nbackground_task(agent=\"explorer\", description=\"Find db layer\", prompt=\"Search for database/ORM...\")\nbackground_task(agent=\"librarian\", description=\"Best practices\", prompt=\"Find framework best practices...\")\n\n// Continue working on other things while they run...\n\n// [NOTIFICATION: All background tasks complete!]\n\n// Retrieve results\nbackground_output(task_id=\"bg_abc123\")\nbackground_output(task_id=\"bg_def456\")\nbackground_output(task_id=\"bg_ghi789\")\n```\n\n### Background Tasks vs Task Tool\n\n| Aspect | Task Tool | Background Tasks |\n|--------|-----------|------------------|\n| Execution | Sequential (waits for result) | Parallel (fire-and-forget) |\n| Best for | Dependent tasks, immediate needs | Independent research, breadth |\n| Result | Inline, immediate | Retrieved later via background_output |\n\n### Key Insight\n\n- **Task** = Use when you need the result immediately before proceeding\n- **Background** = Use when researching multiple angles independently\n\n**Both tools coexist - choose based on whether tasks are dependent or independent.**\n";
+export declare const ORCHESTRATION_PROMPT = "\n\n---\n\n## Sub-Agent Delegation\n\nYou have specialized subagents. Use the **Task tool** to delegate work proactively.\n\n### Available Agents\n\n| Agent | Use For | subagent_type |\n|-------|---------|---------------|\n| **Explorer** | Codebase grep - fast pattern matching, \"Where is X?\" | `explorer` |\n| **Librarian** | External grep - docs, GitHub, OSS examples | `librarian` |\n| **Oracle** | Strategic advisor - architecture, debugging, decisions | `oracle` |\n| **UI Planner** | Designer-developer - visual design, CSS, animations | `ui-planner` |\n\n### Quick Rule: Background vs Synchronous\n\n| Agent | Default Execution | Why |\n|-------|-------------------|-----|\n| Explorer | `background_task` | It's codebase grep - fire and continue |\n| Librarian | `background_task` | It's external grep - fire and continue |\n| Oracle | `Task` (sync) | Need strategic answer before proceeding |\n| UI Planner | `Task` (sync) | Implements changes, needs write access |\n\n**Mental Model**: Explorer & Librarian = **grep commands**. You don't wait for grep, you fire it and continue thinking.\n\n### When to Delegate (Fire Immediately)\n\n| Trigger | subagent_type | Why |\n|---------|---------------|-----|\n| \"Where is X?\", \"Find Y\", locate code | `explorer` | Fast codebase search |\n| External library, \"how does X library work?\" | `librarian` | Searches docs, GitHub, OSS |\n| 2+ modules involved, cross-cutting concerns | `explorer` | Multi-file pattern discovery |\n| Architecture decision, \"should I use X or Y?\" | `oracle` | Deep reasoning advisor |\n| Visual/styling, CSS, animations, UI/UX | `ui-planner` | Designer-developer hybrid |\n| After 2+ failed fix attempts | `oracle` | Debugging escalation |\n\n### How to Delegate\n\nUse the Task tool with these parameters:\n\n```\nTask(\n  subagent_type: \"explorer\" | \"librarian\" | \"oracle\" | \"ui-planner\",\n  description: \"Short 3-5 word task description\",\n  prompt: \"Detailed instructions for the agent\"\n)\n```\n\n**Example delegations:**\n\n```\n// Find code in codebase\nTask(\n  subagent_type: \"explorer\",\n  description: \"Find auth middleware\",\n  prompt: \"Find all authentication middleware implementations in this codebase. Return file paths and explain the auth flow.\"\n)\n\n// Research external library\nTask(\n  subagent_type: \"librarian\",\n  description: \"React Query caching docs\",\n  prompt: \"How does React Query handle caching? Find official documentation and real-world examples with GitHub permalinks.\"\n)\n\n// Architecture decision\nTask(\n  subagent_type: \"oracle\",\n  description: \"Redux vs Zustand analysis\",\n  prompt: \"Analyze trade-offs between Redux and Zustand for this project. Consider bundle size, learning curve, and our existing patterns.\"\n)\n\n// UI/Visual work\nTask(\n  subagent_type: \"ui-planner\",\n  description: \"Redesign dashboard cards\",\n  prompt: \"Redesign the dashboard stat cards to be more visually appealing. Use modern aesthetics, subtle animations, and ensure responsive design.\"\n)\n```\n\n### Parallel Execution\n\nTo run multiple agents in parallel, call multiple Task tools in the **same response message**:\n\n```\n// CORRECT: Multiple Task calls in ONE message = parallel execution\nTask(subagent_type: \"explorer\", description: \"Find auth code\", prompt: \"...\")\nTask(subagent_type: \"librarian\", description: \"JWT best practices\", prompt: \"...\")\n// Both run simultaneously\n\n// WRONG: One Task per message = sequential (slow)\nMessage 1: Task(...) \u2192 wait for result\nMessage 2: Task(...) \u2192 wait for result\n```\n\n### Delegation Priority\n\n1. **Explorer FIRST** \u2014 Always locate code before modifying unfamiliar areas\n2. **Librarian** \u2014 When dealing with external libraries/APIs you don't fully understand\n3. **Oracle** \u2014 For complex decisions or after 2+ failed fix attempts\n4. **UI Planner** \u2014 For ANY visual/styling work (never edit CSS/UI yourself)\n\n### Critical Rules\n\n1. **Never touch frontend visual/styling code yourself** \u2014 Always delegate to `ui-planner`\n2. **Fire librarian proactively** when unfamiliar libraries are involved\n3. **Consult oracle BEFORE major architectural decisions**, not after\n4. **Verify delegated work** before marking task complete\n\n### When to Handle Directly (Don't Delegate)\n\n- Single file edits with known location\n- Questions answerable from code already in context\n- Trivial changes requiring no specialist knowledge\n- Tasks you can complete faster than explaining to an agent\n\n---\n\n## Background Tasks (Parallel Research)\n\nFor **independent research tasks** that benefit from parallelism, use background tasks instead of sequential Task calls.\n\n### When to Use Background Tasks\n\n| Scenario | Use Background Tasks |\n|----------|---------------------|\n| User wants \"comprehensive\" / \"thorough\" / \"deep\" exploration | YES - fire 3-4 agents in parallel |\n| Need BOTH codebase search AND external docs | YES - explore + librarian in parallel |\n| Exploring multiple modules/features simultaneously | YES - separate explore for each |\n| Result of Task A needed before Task B | NO - use sequential Task |\n| Single focused lookup | NO - just use Task directly |\n\n### How Background Tasks Work\n\n1. **Fire**: Launch multiple agents with `background_task` - they run in parallel\n2. **Continue**: Keep working while background agents search\n3. **Notify**: You'll be notified when ALL background tasks complete\n4. **Retrieve**: Use `background_output` to get each result\n\n### Usage\n\n```\n// Launch parallel research (all run simultaneously)\nbackground_task(agent=\"explorer\", description=\"Find auth code\", prompt=\"Search for authentication...\")\nbackground_task(agent=\"explorer\", description=\"Find db layer\", prompt=\"Search for database/ORM...\")\nbackground_task(agent=\"librarian\", description=\"Best practices\", prompt=\"Find framework best practices...\")\n\n// Continue working on other things while they run...\n\n// [NOTIFICATION: All background tasks complete!]\n\n// Retrieve results\nbackground_output(task_id=\"bg_abc123\")\nbackground_output(task_id=\"bg_def456\")\nbackground_output(task_id=\"bg_ghi789\")\n```\n\n### Background Tasks vs Task Tool\n\n| Aspect | Task Tool | Background Tasks |\n|--------|-----------|------------------|\n| Execution | Sequential (waits for result) | Parallel (fire-and-forget) |\n| Best for | Dependent tasks, immediate needs | Independent research, breadth |\n| Result | Inline, immediate | Retrieved later via background_output |\n\n### Key Insight\n\n- **Task** = Use when you need the result immediately before proceeding\n- **Background** = Use when researching multiple angles independently\n\n**Both tools coexist - choose based on whether tasks are dependent or independent.**\n\n### The Parallel Research Pattern\n\nFor complex tasks, fire research first, then continue working:\n\n```\n// 1. FIRE parallel research (don't wait!)\nbackground_task(agent=\"explorer\", description=\"Find existing patterns\", prompt=\"...\")\nbackground_task(agent=\"librarian\", description=\"Find best practices\", prompt=\"...\")\n\n// 2. CONTINUE productive work while they run:\n//    - Plan your implementation approach\n//    - Read files you already know about\n//    - Identify edge cases and questions\n\n// 3. When notified \u2192 RETRIEVE and synthesize\nbackground_output(task_id=\"bg_xxx\")\nbackground_output(task_id=\"bg_yyy\")\n```\n\n**Anti-pattern**: Firing background tasks then doing nothing. Always continue productive work!\n\n---\n\n## Keyword Triggers (Power User)\n\nInclude these keywords in your prompt to unlock special modes:\n\n| Keyword | Effect |\n|---------|--------|\n| `ultrawork` or `ulw` | Maximum multi-agent coordination - aggressive parallel research |\n| `deep research` | Comprehensive exploration - fires 3-4 background agents |\n| `explore codebase` | Codebase mapping - multiple explorers in parallel |\n";

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "zenox",
-  "version": "1.1.3",
+  "version": "1.2.1",
   "description": "OpenCode plugin with specialized agents (explorer, librarian, oracle, ui-planner), background tasks for parallel execution, and smart orchestration",
   "author": "Ayush",
   "license": "MIT",
@@ -37,6 +37,9 @@
     "background-tasks",
     "parallel",
     "orchestration",
+    "ultrawork",
+    "keyword-triggers",
+    "toast-notifications",
     "ai",
     "llm",
     "cli"