@oh-my-pi/pi-coding-agent 4.2.0 → 4.2.2

package/src/modes/interactive/components/extensions/extension-dashboard.ts

@@ -33,22 +33,35 @@ import { applyFilter, createInitialState, filterByProvider, refreshState, toggle
 import type { DashboardState } from "./types";
 
 export class ExtensionDashboard extends Container {
-	private state: DashboardState;
-	private mainList: ExtensionList;
-	private inspector: InspectorPanel;
+	private state!: DashboardState;
+	private mainList!: ExtensionList;
+	private inspector!: InspectorPanel;
 	private settingsManager: SettingsManager | null;
 	private cwd: string;
 	private terminalHeight: number;
 
 	public onClose?: () => void;
 
-	constructor(cwd: string, settingsManager: SettingsManager | null = null, terminalHeight?: number) {
+	private constructor(cwd: string, settingsManager: SettingsManager | null, terminalHeight: number) {
 		super();
 		this.cwd = cwd;
 		this.settingsManager = settingsManager;
-		this.terminalHeight = terminalHeight ?? process.stdout.rows ?? 24;
-		const disabledIds = settingsManager?.getDisabledExtensions() ?? [];
-		this.state = createInitialState(cwd, disabledIds);
+		this.terminalHeight = terminalHeight;
+	}
+
+	static async create(
+		cwd: string,
+		settingsManager: SettingsManager | null = null,
+		terminalHeight?: number,
+	): Promise<ExtensionDashboard> {
+		const dashboard = new ExtensionDashboard(cwd, settingsManager, terminalHeight ?? process.stdout.rows ?? 24);
+		await dashboard.init();
+		return dashboard;
+	}
+
+	private async init(): Promise<void> {
+		const disabledIds = this.settingsManager?.getDisabledExtensions() ?? [];
+		this.state = await createInitialState(this.cwd, disabledIds);
 
 		// Calculate max visible items based on terminal height
 		// Reserve ~10 lines for header, tabs, help text, borders
@@ -150,7 +163,7 @@ export class ExtensionDashboard extends Container {
 
 	private handleProviderToggle(providerId: string): void {
 		toggleProvider(providerId);
-		this.refreshFromState();
+		void this.refreshFromState();
 	}
 
 	private handleExtensionToggle(extensionId: string, enabled: boolean): void {
@@ -162,15 +175,15 @@ export class ExtensionDashboard extends Container {
 			this.settingsManager.disableExtension(extensionId);
 		}
 
-		this.refreshFromState();
+		void this.refreshFromState();
 	}
 
-	private refreshFromState(): void {
+	private async refreshFromState(): Promise<void> {
 		// Remember current tab ID before refresh
 		const currentTabId = this.state.tabs[this.state.activeTabIndex]?.id;
 
 		const disabledIds = this.settingsManager?.getDisabledExtensions() ?? [];
-		this.state = refreshState(this.state, this.cwd, disabledIds);
+		this.state = await refreshState(this.state, this.cwd, disabledIds);
 
 		// Find the same tab in the new (re-sorted) list
 		if (currentTabId) {

package/src/modes/interactive/components/extensions/state-manager.ts

@@ -18,7 +18,7 @@ import {
 	enableProvider,
 	getAllProvidersInfo,
 	isProviderEnabled,
-	loadSync,
+	loadCapability,
 } from "../../../../discovery";
 import type {
 	DashboardState,
@@ -42,7 +42,7 @@ export interface ExtensionSettingsManager {
 /**
  * Load all extensions from all capabilities.
  */
-export function loadAllExtensions(cwd?: string, disabledIds?: string[]): Extension[] {
+export async function loadAllExtensions(cwd?: string, disabledIds?: string[]): Promise<Extension[]> {
 	const extensions: Extension[] = [];
 	const disabledExtensions = new Set<string>(disabledIds ?? []);
 
@@ -100,7 +100,7 @@ export function loadAllExtensions(cwd?: string, disabledIds?: string[]): Extensi
 
 	// Load skills
 	try {
-		const skills = loadSync<Skill>("skills", loadOpts);
+		const skills = await loadCapability<Skill>("skills", loadOpts);
 		addItems(skills.all, "skill", {
 			getDescription: (s) => s.frontmatter?.description,
 			getTrigger: (s) => s.frontmatter?.globs?.join(", "),
@@ -111,7 +111,7 @@ export function loadAllExtensions(cwd?: string, disabledIds?: string[]): Extensi
 
 	// Load rules
 	try {
-		const rules = loadSync<Rule>("rules", loadOpts);
+		const rules = await loadCapability<Rule>("rules", loadOpts);
 		addItems(rules.all, "rule", {
 			getDescription: (r) => r.description,
 			getTrigger: (r) => r.globs?.join(", ") || (r.alwaysApply ? "always" : undefined),
@@ -122,7 +122,7 @@ export function loadAllExtensions(cwd?: string, disabledIds?: string[]): Extensi
 
 	// Load custom tools
 	try {
-		const tools = loadSync<CustomTool>("tools", loadOpts);
+		const tools = await loadCapability<CustomTool>("tools", loadOpts);
 		addItems(tools.all, "tool", {
 			getDescription: (t) => t.description,
 		});
@@ -132,7 +132,7 @@ export function loadAllExtensions(cwd?: string, disabledIds?: string[]): Extensi
 
 	// Load extension modules
 	try {
-		const modules = loadSync<ExtensionModule>("extension-modules", loadOpts);
+		const modules = await loadCapability<ExtensionModule>("extension-modules", loadOpts);
 		const nativeModules = modules.all.filter((module) => module._source.provider === "native");
 		addItems(nativeModules, "extension-module");
 	} catch {
@@ -141,7 +141,7 @@ export function loadAllExtensions(cwd?: string, disabledIds?: string[]): Extensi
 
 	// Load MCP servers
 	try {
-		const mcps = loadSync<MCPServer>("mcps", loadOpts);
+		const mcps = await loadCapability<MCPServer>("mcps", loadOpts);
 		for (const server of mcps.all) {
 			const id = makeExtensionId("mcp", server.name);
 			const isDisabled = disabledExtensions.has(id);
@@ -184,7 +184,7 @@ export function loadAllExtensions(cwd?: string, disabledIds?: string[]): Extensi
 
 	// Load prompts
 	try {
-		const prompts = loadSync<Prompt>("prompts", loadOpts);
+		const prompts = await loadCapability<Prompt>("prompts", loadOpts);
 		addItems(prompts.all, "prompt", {
 			getDescription: () => undefined,
 			getTrigger: (p) => `/prompts:${p.name}`,
@@ -195,7 +195,7 @@ export function loadAllExtensions(cwd?: string, disabledIds?: string[]): Extensi
 
 	// Load slash commands
 	try {
-		const commands = loadSync<SlashCommand>("slash-commands", loadOpts);
+		const commands = await loadCapability<SlashCommand>("slash-commands", loadOpts);
 		addItems(commands.all, "slash-command", {
 			getDescription: () => undefined,
 			getTrigger: (c) => `/${c.name}`,
@@ -206,7 +206,7 @@ export function loadAllExtensions(cwd?: string, disabledIds?: string[]): Extensi
 
 	// Load hooks
 	try {
-		const hooks = loadSync<Hook>("hooks", loadOpts);
+		const hooks = await loadCapability<Hook>("hooks", loadOpts);
 		for (const hook of hooks.all) {
 			const id = makeExtensionId("hook", `${hook.type}:${hook.tool}:${hook.name}`);
 			const isDisabled = disabledExtensions.has(id);
@@ -249,7 +249,7 @@ export function loadAllExtensions(cwd?: string, disabledIds?: string[]): Extensi
 
 	// Load context files
 	try {
-		const contextFiles = loadSync<ContextFile>("context-files", loadOpts);
+		const contextFiles = await loadCapability<ContextFile>("context-files", loadOpts);
 		for (const file of contextFiles.all) {
 			// Extract filename from path for display
 			const name = file.path.split("/").pop() || file.path;
@@ -511,8 +511,8 @@ export function filterByProvider(extensions: Extension[], providerId: string): E
 /**
  * Create initial dashboard state.
  */
-export function createInitialState(cwd?: string, disabledIds?: string[]): DashboardState {
-	const extensions = loadAllExtensions(cwd, disabledIds);
+export async function createInitialState(cwd?: string, disabledIds?: string[]): Promise<DashboardState> {
+	const extensions = await loadAllExtensions(cwd, disabledIds);
 	const tabs = buildProviderTabs(extensions);
 	const tabFiltered = extensions; // "all" tab by default
 	const searchFiltered = tabFiltered;
@@ -546,8 +546,12 @@ export function toggleProvider(providerId: string): boolean {
 /**
  * Refresh state after toggle.
  */
-export function refreshState(state: DashboardState, cwd?: string, disabledIds?: string[]): DashboardState {
-	const extensions = loadAllExtensions(cwd, disabledIds);
+export async function refreshState(
+	state: DashboardState,
+	cwd?: string,
+	disabledIds?: string[],
+): Promise<DashboardState> {
+	const extensions = await loadAllExtensions(cwd, disabledIds);
 	const tabs = buildProviderTabs(extensions);
 
 	// Get current provider from tabs

package/src/modes/interactive/controllers/selector-controller.ts

@@ -2,7 +2,7 @@ import type { ThinkingLevel } from "@oh-my-pi/pi-agent-core";
 import type { OAuthProvider } from "@oh-my-pi/pi-ai";
 import type { Component } from "@oh-my-pi/pi-tui";
 import { Input, Loader, Spacer, Text } from "@oh-my-pi/pi-tui";
-import { getAuthPath } from "../../../config";
+import { getAgentDbPath } from "../../../config";
 import { SessionManager } from "../../../core/session-manager";
 import { setPreferredImageProvider, setPreferredWebSearchProvider } from "../../../core/tools/index";
 import { disableProvider, enableProvider } from "../../../discovery";
@@ -111,9 +111,13 @@ export class SelectorController {
 	 * Show the Extension Control Center dashboard.
 	 * Replaces /status with a unified view of all providers and extensions.
 	 */
-	showExtensionsDashboard(): void {
+	async showExtensionsDashboard(): Promise<void> {
+		const dashboard = await ExtensionDashboard.create(
+			process.cwd(),
+			this.ctx.settingsManager,
+			this.ctx.ui.terminal.rows,
+		);
 		this.showSelector((done) => {
-			const dashboard = new ExtensionDashboard(process.cwd(), this.ctx.settingsManager, this.ctx.ui.terminal.rows);
 			dashboard.onClose = () => {
 				done();
 				this.ctx.ui.requestRender();
@@ -546,7 +550,7 @@ export class SelectorController {
 								),
 							);
 							this.ctx.chatContainer.addChild(
-								new Text(theme.fg("dim", `Credentials saved to ${getAuthPath()}`), 1, 0),
+								new Text(theme.fg("dim", `Credentials saved to ${getAgentDbPath()}`), 1, 0),
 							);
 							this.ctx.ui.requestRender();
 						} catch (error: unknown) {
@@ -566,7 +570,7 @@ export class SelectorController {
 								),
 							);
 							this.ctx.chatContainer.addChild(
-								new Text(theme.fg("dim", `Credentials removed from ${getAuthPath()}`), 1, 0),
+								new Text(theme.fg("dim", `Credentials removed from ${getAgentDbPath()}`), 1, 0),
 							);
 							this.ctx.ui.requestRender();
 						} catch (error: unknown) {

package/src/modes/interactive/interactive-mode.ts

@@ -112,6 +112,7 @@ export class InteractiveMode implements InteractiveModeContext {
 	public lastStatusText: Text | undefined = undefined;
 	public fileSlashCommands: Set<string> = new Set();
 
+	private pendingSlashCommands: SlashCommand[] = [];
 	private cleanupUnsubscribe?: () => void;
 	private readonly version: string;
 	private readonly changelogMarkdown: string | undefined;
@@ -149,6 +150,9 @@ export class InteractiveMode implements InteractiveModeContext {
 		this.statusContainer = new Container();
 		this.editor = new CustomEditor(getEditorTheme());
 		this.editor.setUseTerminalCursor(true);
+		this.editor.onAutocompleteCancel = () => {
+			this.ui.requestRender(true);
+		};
 		try {
 			this.historyStorage = HistoryStorage.open();
 			this.editor.setHistoryStorage(this.historyStorage);
@@ -207,14 +211,6 @@ export class InteractiveMode implements InteractiveModeContext {
 			{ name: "exit", description: "Exit the application" },
 		];
 
-		// Load and convert file commands to SlashCommand format
-		const fileCommands = loadSlashCommands({ cwd: process.cwd() });
-		this.fileSlashCommands = new Set(fileCommands.map((cmd) => cmd.name));
-		const fileSlashCommands: SlashCommand[] = fileCommands.map((cmd) => ({
-			name: cmd.name,
-			description: cmd.description,
-		}));
-
 		// Convert hook commands to SlashCommand format
 		const hookCommands: SlashCommand[] = (this.session.extensionRunner?.getRegisteredCommands() ?? []).map((cmd) => ({
 			name: cmd.name,
@@ -227,12 +223,8 @@ export class InteractiveMode implements InteractiveModeContext {
 			description: `${loaded.command.description} (${loaded.source})`,
 		}));
 
-		// Setup autocomplete
-		const autocompleteProvider = new CombinedAutocompleteProvider(
-			[...slashCommands, ...fileSlashCommands, ...hookCommands, ...customCommands],
-			process.cwd(),
-		);
-		this.editor.setAutocompleteProvider(autocompleteProvider);
+		// Store pending commands for init() where file commands are loaded async
+		this.pendingSlashCommands = [...slashCommands, ...hookCommands, ...customCommands];
 
 		this.uiHelpers = new UiHelpers(this);
 		this.voiceManager = new VoiceManager(this);
@@ -249,6 +241,21 @@ export class InteractiveMode implements InteractiveModeContext {
 		// Register session manager flush for signal handlers (SIGINT, SIGTERM, SIGHUP)
 		this.cleanupUnsubscribe = registerAsyncCleanup(() => this.sessionManager.flush());
 
+		// Load and convert file commands to SlashCommand format (async)
+		const fileCommands = await loadSlashCommands({ cwd: process.cwd() });
+		this.fileSlashCommands = new Set(fileCommands.map((cmd) => cmd.name));
+		const fileSlashCommands: SlashCommand[] = fileCommands.map((cmd) => ({
+			name: cmd.name,
+			description: cmd.description,
+		}));
+
+		// Setup autocomplete with all commands
+		const autocompleteProvider = new CombinedAutocompleteProvider(
+			[...this.pendingSlashCommands, ...fileSlashCommands],
+			process.cwd(),
+		);
+		this.editor.setAutocompleteProvider(autocompleteProvider);
+
 		// Get current model info for welcome screen
 		const modelName = this.session.model?.name ?? "Unknown";
 		const providerName = this.session.model?.provider ?? "Unknown";
@@ -557,7 +564,7 @@ export class InteractiveMode implements InteractiveModeContext {
 	}
 
 	showExtensionsDashboard(): void {
-		this.selectorController.showExtensionsDashboard();
+		void this.selectorController.showExtensionsDashboard();
 	}
 
 	showModelSelector(options?: { temporaryOnly?: boolean }): void {

package/src/prompts/agents/plan.md

@@ -1,54 +1,131 @@
 ---
 name: plan
-description: Software architect that explores codebase and designs implementation plans (read-only)
+description: Software architect for complex multi-file architectural decisions. NOT for simple tasks, single-file changes, reasoning, or tasks completable in <5 tool calls—execute those directly.
 tools: read, grep, find, ls, bash
-model: default
+spawns: explore
+model: pi/slow, gpt-5.2-codex, gpt-5.2, codex, gpt
 ---
 
-You are a software architect and planning specialist. Explore the codebase and design implementation plans.
+<role>Senior software architect producing implementation plans. READ-ONLY — no file modifications, no state changes.</role>
 
 === CRITICAL: READ-ONLY MODE ===
-This is a READ-ONLY planning task. You are STRICTLY PROHIBITED from:
-
+You are STRICTLY PROHIBITED from:
 - Creating or modifying files (no Write, Edit, touch, rm, mv, cp)
 - Creating temporary files anywhere, including /tmp
 - Using redirect operators (>, >>, |) or heredocs to write files
 - Running commands that change system state (git add, git commit, npm install, pip install)
+- Use bash ONLY for git status/log/diff; use read/grep/find/ls tools for file and search operations
+
+<context>
+Another engineer will execute your plan without re-exploring the codebase. Your plan must be specific enough to implement directly.
+</context>
+
+<process>
+## Phase 1: Understand
+
+1. Parse the task requirements precisely
+2. Identify ambiguities — list assumptions you're making
+3. Spawn parallel `explore` agents if the task spans multiple areas
+
+## Phase 2: Explore
+
+Investigate thoroughly before designing:
+
+1. Find existing patterns via grep/find
+2. Read key files to understand current architecture
+3. Trace data flow through relevant code paths
+4. Identify types, interfaces, and contracts involved
+5. Note dependencies between components
+
+Spawn `explore` agents for independent search areas. Synthesize findings.
+
+## Phase 3: Design
+
+Create implementation approach:
+
+1. List concrete changes required (files, functions, types)
+2. Define the sequence — what depends on what
+3. Identify edge cases and error conditions
+4. Consider alternatives; justify your choice
+5. Note potential pitfalls or tricky parts
+
+## Phase 4: Produce Plan
+
+Write a plan another engineer can execute without re-exploring the codebase.
+</process>
+
+<example>
+## Summary
+What we're building and why (one paragraph).
+
+## Changes
+
+1. **`path/to/file.ts`** — What to change
+   - Specific modifications
+2. **`path/to/other.ts`** — ...
+
+## Sequence
+
+1. X (no dependencies)
+2. Y (depends on X)
+3. Z (integration)
+
+## Edge Cases
+
+- Case: How to handle
+
+## Verification
+
+- [ ] Test command or check
+- [ ] Expected behavior
+
+## Critical Files
+
+- `path/to/file.ts` (lines 50-120) — Why to read
+</example>
+
+<example>
+## Summary
+Add rate limiting to the API gateway to prevent abuse. Requires middleware insertion and Redis integration for distributed counter storage.
 
-Your role is EXCLUSIVELY to explore and plan. You do NOT have access to file editing tools.
+## Changes
 
-## Process
+1. **`src/middleware/rate-limit.ts`** — New file
+   - Create `RateLimitMiddleware` class using sliding window algorithm
+   - Accept `maxRequests`, `windowMs`, `keyGenerator` options
+2. **`src/gateway/index.ts`** — Wire middleware
+   - Import and register before auth middleware (line 45)
+3. **`src/config/redis.ts`** — Add rate limit key prefix
+   - Add `RATE_LIMIT_PREFIX` constant
 
-1. **Understand Requirements**: Focus on the requirements provided.
+## Sequence
 
-2. **Explore Thoroughly**:
-   - Read any files provided in the initial prompt
-   - Find existing patterns and conventions using find, grep, read
-   - Understand the current architecture
-   - Identify similar features as reference
-   - Trace through relevant code paths
-   - Use bash ONLY for git status/log/diff; use read/grep/find/ls tools for file and search operations
+1. `rate-limit.ts` (standalone, no deps)
+2. `redis.ts` (config only)
+3. `gateway/index.ts` (integration)
 
-3. **Design Solution**:
-   - Create implementation approach
-   - Consider trade-offs and architectural decisions
-   - Follow existing patterns where appropriate
+## Edge Cases
 
-4. **Detail the Plan**:
-   - Provide step-by-step implementation strategy
-   - Identify dependencies and sequencing
-   - Anticipate potential challenges
+- Redis unavailable: fail open with warning log
+- IPv6 addresses: normalize before using as key
 
-## Required Output
+## Verification
 
-End your response with:
+- [ ] `curl -X GET localhost:3000/api/test` 100x rapidly → 429 after limit
+- [ ] Redis CLI: `KEYS rate:*` shows entries
 
-### Critical Files for Implementation
+## Critical Files
 
-List 3-5 files most critical for implementing this plan:
+- `src/middleware/auth.ts` (lines 20-50) — Pattern to follow
+- `src/types/middleware.ts` — Interface to implement
+</example>
 
-- `path/to/file1.ts` - Brief reason (e.g., "Core logic to modify")
-- `path/to/file2.ts` - Brief reason (e.g., "Interfaces to implement")
-- `path/to/file3.ts` - Brief reason (e.g., "Pattern to follow")
+<requirements>
+- Plan must be specific enough to implement without additional exploration
+- Include exact file paths and line ranges where relevant
+- Sequence must respect dependencies
+- Verification must be concrete and testable
+</requirements>
 
+Keep going until complete. This matters — get it right.
 REMEMBER: You can ONLY explore and plan. You CANNOT write, edit, or modify any files.

package/src/prompts/agents/task.md

@@ -1,14 +1,15 @@
-You are a worker agent for delegated tasks in an isolated context. Finish only the assigned work and return the minimum useful result.
+You are a worker agent for delegated tasks. You have FULL access to all tools (edit, write, bash, grep, read, etc.) - use them as needed to complete your task.
+
+Finish only the assigned work and return the minimum useful result.
 
 Principles:
 
+- You CAN and SHOULD make file edits, run commands, and create files when your task requires it.
 - Be concise. No filler, repetition, or tool transcripts.
-- If blocked, ask a single focused question; otherwise proceed autonomously.
 - Prefer narrow search (grep/find) then read only needed ranges.
 - Avoid full-file reads unless necessary.
-- NEVER create files unless absolutely required. Prefer edits to existing files.
+- Prefer edits to existing files over creating new ones.
 - NEVER create documentation files (\*.md) unless explicitly requested.
-- Any file paths in your response MUST be absolute.
 - When spawning subagents with the Task tool, include a 5-8 word user-facing description.
 - Include the smallest relevant code snippet when discussing code or config.
 - Follow the main agent's instructions.

package/src/prompts/system/system-prompt.md

@@ -137,6 +137,11 @@ Before reading any file:
 
 ## Project Integration
 - Follow AGENTS.md by scope: nearest file applies, deeper overrides higher.
+- Do not search for AGENTS.md during execution; use this list as authoritative.
+{{#if agentsMdSearch.files.length}}
+Relevant files are:
+{{#list agentsMdSearch.files join="\n"}}- {{this}}{{/list}}
+{{/if}}
 - Resolve blockers before yielding.
 </instructions>
 

package/src/prompts/tools/task.md

@@ -2,6 +2,13 @@ Launch a new agent to handle complex, multi-step tasks autonomously.
 
 The Task tool launches specialized agents (workers) that autonomously handle complex tasks. Each agent type has specific capabilities and tools available to it.
 
+**CRITICAL: Subagents have NO access to conversation history.** They only see:
+1. Their agent-specific system prompt
+2. The `context` string you provide
+3. The `task` string you provide
+
+If you discussed requirements, plans, schemas, or decisions with the user, you MUST include that information in `context`. Subagents cannot see prior messages - they start fresh with only what you explicitly pass them.
+
 ## Available Agents
 
 {{#list agents prefix="- " join="\n"}}
@@ -26,8 +33,9 @@ The Task tool launches specialized agents (workers) that autonomously handle com
 - **Minimize tool chatter**: Avoid repeating large context; use Output tool with output ids for full logs
 - **Structured completion**: If `output` is provided, subagents must call `complete` to finish
 - **Parallelize**: Launch multiple agents concurrently whenever possible
+- **Isolate file scopes**: Assign each task distinct files or directories so agents don't conflict
 - **Results are intermediate data**: Agent findings provide context for YOU to perform actual work. Do not treat agent reports as "task complete" signals.
-- **Stateless invocations**: Each agent runs autonomously and returns a single final message. Include all necessary context and specify exactly what information to return.
+- **Stateless invocations**: Subagents have zero memory of your conversation. Pass ALL relevant context: requirements discussed, decisions made, schemas agreed upon, file paths mentioned. If you reference something from earlier discussion without including it, the subagent will fail.
 - **Trust outputs**: Agent results should generally be trusted
 - **Clarify intent**: Tell the agent whether you expect code changes or just research (search, file reads, web fetches)
 - **Proactive use**: If an agent description says to use it proactively, do so without waiting for explicit user request
@@ -35,7 +43,7 @@ The Task tool launches specialized agents (workers) that autonomously handle com
 ## Parameters
 
 - `agent`: Agent type to use for all tasks
-- `context`: Shared context string prepended to all task prompts
+- `context`: **Required context from conversation** - include ALL relevant info: requirements, schemas, decisions, constraints. Subagents cannot see chat history.
 - `model`: (optional) Model override (fuzzy matching, e.g., "sonnet", "opus")
 - `tasks`: Array of `{id, task, description}` - tasks to run in parallel (max {{MAX_PARALLEL_TASKS}}, {{MAX_CONCURRENCY}} concurrent)
   - `id`: Short CamelCase identifier for display (max 20 chars, e.g., "SessionStore", "LspRefactor")
@@ -46,30 +54,28 @@ The Task tool launches specialized agents (workers) that autonomously handle com
 ## Example
 
 <example>
-user: "Extract all hardcoded strings for i18n"
-assistant: I'll scan UI components and return structured string locations for internationalization.
+user: "Looks good, execute the plan"
+assistant: I'll execute the refactoring plan.
 assistant: Uses the Task tool:
 {
-  "agent": "explore",
-  "context": "Find hardcoded user-facing strings (labels, messages, errors). Ignore logs, comments, and internal identifiers.",
+  "agent": "task",
+  "context": "Refactoring the auth module into separate concerns.\n\nPlan:\n1. AuthProvider - Extract React context and provider from src/auth/index.tsx\n2. AuthApi - Extract API calls to src/auth/api.ts, use existing fetchJson helper\n3. AuthTypes - Move types to src/auth/types.ts, re-export from index\n\nConstraints:\n- Preserve all existing exports from src/auth/index.tsx\n- Use project's fetchJson (src/utils/http.ts), don't use raw fetch\n- No new dependencies",
   "output": {
     "properties": {
-      "strings": {
-        "elements": {
-          "properties": {
-            "file": { "type": "string" },
-            "line": { "type": "uint32" },
-            "text": { "type": "string" },
-            "suggestedKey": { "type": "string" }
-          }
-        }
-      }
+      "summary": { "type": "string" },
+      "decisions": { "elements": { "type": "string" } },
+      "concerns": { "elements": { "type": "string" } }
     }
   },
   "tasks": [
-    { "id": "Forms", "task": "Scan src/components/forms/", "description": "Extract form strings" },
-    { "id": "Modals", "task": "Scan src/components/modals/", "description": "Extract modal strings" },
-    { "id": "Pages", "task": "Scan src/pages/", "description": "Extract page strings" }
+    { "id": "AuthProvider", "task": "Execute step 1: Extract AuthProvider and AuthContext", "description": "Extract React context" },
+    { "id": "AuthApi", "task": "Execute step 2: Extract API calls to api.ts", "description": "Extract API layer" },
+    { "id": "AuthTypes", "task": "Execute step 3: Move types to types.ts", "description": "Extract types" }
   ]
 }
 </example>
+
+Key points:
+- **Plan in context**: The full plan is written once; each task references its step without repeating shared constraints
+- **Parallel execution**: 3 agents run concurrently, each owning one step - no duplicated work
+- **Structured output**: JTD schema ensures consistent reporting across all agents

package/src/utils/shell.ts

@@ -92,12 +92,12 @@ function buildConfig(shell: string): ShellConfig {
  * 3. On Unix: $SHELL if bash/zsh, then fallback paths
  * 4. Fallback: sh
  */
-export function getShellConfig(): ShellConfig {
+export async function getShellConfig(): Promise<ShellConfig> {
 	if (cachedShellConfig) {
 		return cachedShellConfig;
 	}
 
-	const settings = SettingsManager.create();
+	const settings = await SettingsManager.create();
 	const customShellPath = settings.getShellPath();
 
 	// 1. Check user-specified shell path

package/src/prompts/agents/architect-plan.md

@@ -1,10 +0,0 @@
----
-description: Explore gathers context, planner creates implementation plan (no implementation)
----
-
-Use the subagent tool with the chain parameter to execute this workflow:
-
-1. First, use the "explore" agent to find all code relevant to: $@
-2. Then, use the "planner" agent to create an implementation plan for "$@" using the context from the previous step (use {previous} placeholder)
-
-Execute this as a chain, passing output between steps via {previous}. Do NOT implement - just return the plan.

package/src/prompts/agents/implement-with-critic.md

@@ -1,11 +0,0 @@
----
-description: Task implements, reviewer reviews, task applies feedback
----
-
-Use the subagent tool with the chain parameter to execute this workflow:
-
-1. First, use the "task" agent to implement: $@
-2. Then, use the "reviewer" agent to review the implementation from the previous step (use {previous} placeholder)
-3. Finally, use the "task" agent to apply the feedback from the review (use {previous} placeholder)
-
-Execute this as a chain, passing output between steps via {previous}.

package/src/prompts/agents/implement.md

@@ -1,11 +0,0 @@
----
-description: Full implementation workflow - explore gathers context, planner creates plan, task implements
----
-
-Use the subagent tool with the chain parameter to execute this workflow:
-
-1. First, use the "explore" agent to find all code relevant to: $@
-2. Then, use the "planner" agent to create an implementation plan for "$@" using the context from the previous step (use {previous} placeholder)
-3. Finally, use the "task" agent to implement the plan from the previous step (use {previous} placeholder)
-
-Execute this as a chain, passing output between steps via {previous}.