taskplane 0.24.31 → 0.25.0

package/README.md

@@ -1,35 +1,47 @@
 # Taskplane
 
-Multi-agent AI orchestration for [pi](https://github.com/badlogic/pi-mono) — parallel task execution with checkpoint discipline, fresh-context worker loops, cross-model reviews, and automated merges.
+Multi-agent AI orchestration for coding with [pi](https://github.com/badlogic/pi-mono) — parallel task execution, mono- and poly-repo support, fresh-context worker loops, cross-model reviews, automated merges and a killer dashboard! 
 
-> **Status:** Experimental / Early — APIs and config formats may change between releases.
+> **Status:** Initial release.
 
 ## What It Does
 
-### STEP 1: Create the tasks
-Taskplane turns your coding project into an AI-managed task orchestration system. You simply ask your agent to create tasks using the built-in "create-taskplane-tasks" skill. This skill provides an opinionated task definition template designed to drive successful coding outcomes. Tasks define both the prompt.md and the status.md files that together act as the persistent memory store that allows AI coding agents to survive context resets and succeed with very long running tasks that would typically exhaust an agent's context window.
+Taskplane turns ideas into high-quality code using a proven process of:
 
-### STEP 2: Run batches of tasks
-The system works out the dependancy map for the entire batch of tasks then orchestrates them in waves, with appropriate parallelization and serialization. 
+have an idea >> create a spec >> create tasks >> orchestrate tasks >> evaluate the outcome
 
-The taskplane dashboard runs on a local port on your system and gives you elegant visibility into everything that's going on (a stark improvement over TUI-based dashboards).
+### Taskplane has:
+- A skill for creating tasks that the Taskplane orchestrator can run
+- PROMPT.md/STATUS.md task definition for persistent memory store
+- Support for both monorepo and polyrepo projects
+- Complete parallelized worktree isolation with dependency graphing and segment-level repo isolation
+- 4 agent types: supervisor, worker, reviewer, and merger
+- A deterministic orchestration engine to drive repeatable positive agent outcomes at scale
+- A simple file-based mail system so agents can communicate with each other
+- A killer locally-run web-based dashboard so you can see everything that's going on
 
 <img src="docs/images/orchrun-wave2of4-2lanes-withstatus.png" alt="image of taskplane dashboard" width="50%">
 
+### STEP 1: Create the tasks
+Taskplane turns your coding project into an AI-managed task orchestration system. You simply ask your agent to create tasks using the built-in "create-taskplane-tasks" skill. This skill provides an opinionated task definition template designed to drive successful coding outcomes. Tasks define both the prompt.md and the status.md files that together act as the persistent memory store that allows AI coding agents to survive context resets and succeed with very long running tasks that would typically exhaust an agent's context window.
+
+### STEP 2: Run batches of tasks
+Taskplane works out the dependency map for an entire batch of tasks then orchestrates them in waves, lanes, and tasks with appropriate parallelization and serialization. Taskplane can do this for both monorepo and polyrepo projects. For polyrepo projects, Taskplane additionally subdivides tasks into repo-aligned segments and uses a segmentation dependency map (DAG) to manage proper repo/worktree isolation and allow for dynamic segment expansion so worker agents can ask the supervisor agent to add additional segments to the dependency map in real time if required.
+
 ### Key Features
 
 - **Task Orchestrator** — Parallel multi-task execution using git worktrees for full filesystem isolation. Dependency-aware wave scheduling. Automated merges into a dedicated orch branch — your working branch stays stable until you choose to integrate.
-- **Persistent Worker Context** — Workers handle all steps in a single context, auto-detecting the model's context window (1M for Claude 4.6 Opus, 200K for Bedrock). Only iterates on context overflow. Dramatic reduction in spawn count and token cost.
-- **Worker-Driven Inline Reviews** — Workers invoke a `review_step` tool at step boundaries. Reviewer agents spawn in tmux sessions with full telemetry. REVISE feedback is addressed inline without losing context.
+- **Persistent Worker Context** — Workers handle all steps in a single context, auto-detecting the model's context window. Only iterates on context overflow. Dramatic reduction in spawn count and token cost.
+- **Worker-Driven Inline Reviews** — Workers invoke a `review_step` tool at step boundaries. Reviewer agents spawn with full telemetry. REVISE feedback is addressed inline without losing context.
 - **Supervisor Agent** — Conversational supervisor monitors batch progress, handles failures, and can invoke orchestrator commands autonomously (resume, integrate, pause, abort).
 - **Web Dashboard** — Live browser-based monitoring via `taskplane dashboard`. SSE streaming, lane/task progress, reviewer activity, merge telemetry, batch history.
 - **Structured Tasks** — PROMPT.md defines the mission, steps, and constraints. STATUS.md tracks progress. Agents follow the plan, not vibes.
 - **Checkpoint Discipline** — Step boundary commits ensure work is never lost, even if a worker crashes mid-task.
 - **Cross-Model Review** — Reviewer agent uses a different model than the worker agent (highly recommended, not enforced). Independent quality gate before merge.
 
-## Install
+## Installation
 
-Taskplane is a [pi package](https://github.com/badlogic/pi-mono). You need [Node.js](https://nodejs.org/) ≥ 22 and [pi](https://github.com/badlogic/pi-mono) installed first.
+Taskplane is a pi package. You need Node.js 22+, pi and Git installed first.
 
 ### Prerequisites
 
@@ -38,51 +50,31 @@ Taskplane is a [pi package](https://github.com/badlogic/pi-mono). You need [Node
 | [Node.js](https://nodejs.org/) ≥ 22 | Yes | Runtime |
 | [pi](https://github.com/badlogic/pi-mono) | Yes | Agent framework |
 | [Git](https://git-scm.com/) | Yes | Version control, worktrees |
-| **tmux** | **Strongly recommended** | Required for `/orch` parallel execution |
 
-**tmux** is needed for the orchestrator to spawn parallel worker sessions. Without it, `/orch` will not work. On Windows, Taskplane can install it for you:
-
-```bash
-taskplane install-tmux
-```
-
-On macOS: `brew install tmux` · On Linux: `sudo apt install tmux` (or your distro's package manager)
-
-### Option A: Global Install (all projects)
+### Option A: Global Install (all projects - recommended)
 
 ```bash
 pi install npm:taskplane
 ```
 
-### Option B: Project-Local Install (recommended for teams)
+### Option B: Single Project-Local Install
 
 ```bash
 cd my-project
 pi install -l npm:taskplane
 ```
 
-Then scaffold your project:
-
-```bash
-taskplane init
-```
-
-Verify the installation:
-
-```bash
-taskplane doctor
-```
-
 ## Quickstart
 
-### 1. Initialize a project
+### 1. Initialize a project (scaffolds settings)
 
 ```bash
 cd my-project
-taskplane init --preset full
+taskplane init
 ```
+You'll answer a few questions. You can usually just accept the defaults. 
 
-This creates config files in `.pi/`, agent prompts, two example tasks, and adds `.gitignore` entries for runtime artifacts. On first install, init bootstraps global preferences at `~/.pi/agent/taskplane/preferences.json` with thinking defaults set to `high` for worker/reviewer/merger. Interactive init then prompts for worker/reviewer/merger model + thinking defaults (`inherit`, `off`, `minimal`, `low`, `medium`, `high`, `xhigh`). If 2+ providers are available from `pi --list-models`, init recommends cross-provider reviewer/merger selections. Init auto-detects whether you're in a single repo or a multi-repo workspace. See the [install tutorial](docs/tutorials/install.md) for workspace mode and other scenarios.
+This creates config files in `.pi/`, agent prompts, two example tasks, and adds `.gitignore` entries for runtime artifacts. On first install, init bootstraps global preferences at `~/.pi/agent/taskplane/preferences.json` with thinking defaults set to `high` for worker & reviewer, and off for merger. Interactive init then prompts for worker/reviewer/merger model + thinking defaults (`inherit`, `off`, `minimal`, `low`, `medium`, `high`, `xhigh`). If 2+ providers are available from `pi --list-models`, init recommends cross-provider reviewer/merger selections. Init auto-detects whether you're in a single repo or a multi-repo workspace. See the [install tutorial](docs/tutorials/install.md) for workspace mode and other scenarios.
 
 Want to reuse model/thinking picks across projects? Run `taskplane config --save-as-defaults` in an initialized project.
 
@@ -94,7 +86,15 @@ taskplane init --preset full --tasks-root docs/task-management
 
 When `--tasks-root` is provided, example task packets are skipped by default. Add `--include-examples` if you explicitly want examples in that folder.
 
-### 2. Launch the dashboard (recommended)
+### 2. Check your install with taskplane doctor
+
+Verify the installation and scaffolding. You should have all green checkboxes if everything was successful:
+
+```bash
+taskplane doctor
+```
+
+### 3. Launch the dashboard (recommended)
 
 In a separate terminal:
 
@@ -104,7 +104,7 @@ taskplane dashboard
 
 Opens a live web dashboard at `http://localhost:8099` with real-time batch monitoring.
 
-### 3. Run your first orchestration
+### 4. Run your first orchestration
 
 ```bash
 pi
@@ -121,7 +121,7 @@ Inside the pi session:
 
 `/orch` with no arguments is the universal entry point — it detects your project state and activates the supervisor for guided interaction (onboarding, batch planning, health checks, or retrospective). The default scaffold includes two independent example tasks, so `/orch all` gives you an immediate orchestrator + dashboard experience.
 
-### 4. Run a single task with isolation
+### 5. Run a single task with isolation
 
 For a single task with full worktree isolation, dashboard, and reviews:
 
@@ -131,18 +131,12 @@ For a single task with full worktree isolation, dashboard, and reviews:
 
 This uses the same orchestrator infrastructure as a full batch — isolated worktree, orch branch, supervisor, dashboard, inline reviews — but for just one task.
 
-> **Deprecated:** The `/task` command is deprecated and will be removed in a future major version. It does not provide worktree isolation, dashboard, or inline reviews. Use `/orch` for all workflows — including single-task execution.
-
 ## Commands
 
 ### Pi Session Commands
 
 | Command | Description |
 |---------|-------------|
-| `/task <path/to/PROMPT.md>` | ⚠️ **Deprecated.** Execute one task in the current branch/worktree. Use `/orch` instead. |
-| `/task-status` | ⚠️ **Deprecated.** Show current task progress. Use `/orch-status` or dashboard. |
-| `/task-pause` | ⚠️ **Deprecated.** Pause after current worker iteration finishes. Use `/orch-pause`. |
-| `/task-resume` | ⚠️ **Deprecated.** Resume a paused task. Use `/orch-resume`. |
 | `/orch [<areas\|paths\|all>]` | No args: detect state & guide (onboarding, batch planning, etc.); with args: execute tasks via isolated worktrees |
 | `/orch-plan <areas\|paths\|all>` | Preview execution plan without running |
 | `/orch-status` | Show batch progress |
@@ -176,8 +170,7 @@ This uses the same orchestrator infrastructure as a full batch — isolated work
        │          │          │
   ┌────▼────┐ ┌──▼─────┐ ┌──▼─────┐
   │ Lane 1  │ │ Lane 2 │ │ Lane 3 │    ← Git worktrees
-  │ /task   │ │ /task  │ │ /task  │       (isolated)
-  │ Worker  │ │ Worker │ │ Worker │
+  │ Worker  │ │ Worker │ │ Worker │       (isolated)
   │ Review  │ │ Review │ │ Review │
   └────┬────┘ └──┬─────┘ └──┬─────┘
        │         │          │

package/bin/taskplane.mjs

@@ -316,93 +316,6 @@ function detectStack(projectRoot) {
 
 // ─── YAML Generation ────────────────────────────────────────────────────────
 
-function generateTaskRunnerYaml(vars) {
-	return `# ═══════════════════════════════════════════════════════════════════════
-# Task Runner Configuration — ${vars.project_name}
-# ═══════════════════════════════════════════════════════════════════════
-#
-# This file configures the /task command (task-runner extension).
-# Edit freely — this file is owned by you, not the package.
-
-# ── Task Areas ────────────────────────────────────────────────────────
-# Define where tasks live. Each area has a folder path, ID prefix, and
-# a CONTEXT.md file that provides domain context to agents.
-
-task_areas:
-  ${vars.default_area}:
-    path: "${vars.tasks_root}"
-    prefix: "${vars.default_prefix}"
-    context: "${vars.tasks_root}/CONTEXT.md"
-
-# ── Reference Docs ────────────────────────────────────────────────────
-# Docs that tasks can reference in their "Context to Read First" section.
-# Add your project's architecture docs, API specs, etc.
-
-reference_docs: {}
-
-# ── Standards ─────────────────────────────────────────────────────────
-# Coding standards and rules. Agents follow these during implementation.
-
-standards: {}
-
-# ── Testing ───────────────────────────────────────────────────────────
-# Commands that agents run to verify their work.
-
-testing:
-  commands:${vars.test_cmd ? `\n    unit: "${vars.test_cmd}"` : ""}${vars.build_cmd ? `\n    build: "${vars.build_cmd}"` : ""}
-`;
-}
-
-function generateOrchestratorYaml(vars) {
-	return `# ═══════════════════════════════════════════════════════════════════════
-# Parallel Task Orchestrator Configuration — ${vars.project_name}
-# ═══════════════════════════════════════════════════════════════════════
-#
-# This file configures the /orch commands (task-orchestrator extension).
-# Edit freely — this file is owned by you, not the package.
-
-orchestrator:
-  max_lanes: ${vars.max_lanes}
-  worktree_location: "subdirectory"
-  worktree_prefix: "${vars.worktree_prefix}"
-  batch_id_format: "timestamp"
-  spawn_mode: "${vars.spawn_mode}"
-  session_prefix: "${vars.session_prefix}"
-
-dependencies:
-  source: "prompt"
-  cache: true
-
-assignment:
-  strategy: "affinity-first"
-  size_weights:
-    S: 1
-    M: 2
-    L: 4
-
-pre_warm:
-  auto_detect: false
-  commands: {}
-  always: []
-
-merge:
-  model: ""
-  tools: "read,write,edit,bash,grep,find,ls"
-  verify: []
-  order: "fewest-files-first"
-
-failure:
-  on_task_failure: "skip-dependents"
-  on_merge_failure: "pause"
-  stall_timeout: 30
-  max_worker_minutes: 30
-  abort_grace_period: 60
-
-monitoring:
-  poll_interval: 5
-`;
-}
-
 function buildTestingCommands(vars) {
 	const commands = {};
 	if (vars.test_cmd) commands.unit = vars.test_cmd;
@@ -964,6 +877,27 @@ async function autoCommitTaskFiles(projectRoot, tasksRoot) {
 }
 
 function discoverTaskAreaMetadata(projectRoot, configRoot = projectRoot, configPrefix = ".pi") {
+	// Prefer taskplane-config.json; fall back to task-runner.yaml for legacy projects
+	const jsonPath = path.join(configRoot, configPrefix, "taskplane-config.json");
+	if (fs.existsSync(jsonPath)) {
+		try {
+			const config = JSON.parse(fs.readFileSync(jsonPath, "utf8"));
+			const areas = config?.taskRunner?.taskAreas;
+			if (areas && typeof areas === "object" && !Array.isArray(areas)) {
+				const paths = new Set();
+				const contexts = new Set();
+				const areaRepoIds = {};
+				for (const [areaName, area] of Object.entries(areas)) {
+					if (!area || typeof area !== "object") continue;
+					if (typeof area.path === "string" && area.path) paths.add(area.path);
+					if (typeof area.context === "string" && area.context) contexts.add(area.context);
+					if (typeof area.repoId === "string" && area.repoId) areaRepoIds[areaName] = area.repoId;
+				}
+				return { paths: [...paths], contexts: [...contexts], areaRepoIds };
+			}
+		} catch { /* fall through to YAML */ }
+	}
+
 	const runnerPath = path.join(configRoot, configPrefix, "task-runner.yaml");
 	if (!fs.existsSync(runnerPath)) return { paths: [], contexts: [], areaRepoIds: {} };
 
@@ -1132,9 +1066,10 @@ async function cmdUninstall(args) {
 	console.log(`\n${c.bold}Taskplane Uninstall${c.reset}\n`);
 
 	const managedFiles = [
+		".pi/taskplane-config.json",
+		".pi/taskplane.json",
 		".pi/task-runner.yaml",
 		".pi/task-orchestrator.yaml",
-		".pi/taskplane.json",
 		".pi/agents/task-worker.md",
 		".pi/agents/task-reviewer.md",
 		".pi/agents/task-merger.md",
@@ -1194,7 +1129,7 @@ async function cmdUninstall(args) {
 		for (const f of sidecarsToDelete) console.log(`  - remove ${f.rel}`);
 		for (const d of taskDirsToDelete) console.log(`  - remove dir ${d.rel}`);
 		if (removeTasks && taskDirsToDelete.length === 0) {
-			console.log(`  ${c.dim}No task area directories found from .pi/task-runner.yaml.${c.reset}`);
+			console.log(`  ${c.dim}No task area directories found in config.${c.reset}`);
 		}
 		if (!removeTasks) {
 			console.log(`  ${c.dim}Task directories are preserved by default (use --remove-tasks to delete them).${c.reset}`);
@@ -1637,6 +1572,10 @@ async function cmdInit(args) {
 	const tasksRootIdx = args.indexOf("--tasks-root");
 	const tasksRootRaw = tasksRootIdx !== -1 ? args[tasksRootIdx + 1] : null;
 
+	if (preset && preset !== "minimal" && preset !== "full") {
+		die(`Unknown preset: "${preset}". Valid presets are: minimal, full`);
+	}
+
 	if (noExamplesFlag && includeExamples) {
 		die("Choose either --no-examples or --include-examples, not both.");
 	}
@@ -1670,7 +1609,7 @@ async function cmdInit(args) {
 
 	// ── Mode auto-detection ──────────────────────────────────────
 	const detection = detectInitMode(projectRoot);
-	const isPreset = preset === "minimal" || preset === "full" || preset === "runner-only";
+	const isPreset = preset === "minimal" || preset === "full";
 
 	// Error path: not a git repo and no git repos found
 	if (detection.mode === "error") {
@@ -1911,7 +1850,7 @@ async function cmdInit(args) {
 
 		// ── Gather config values (workspace mode) ───────────────────
 		let vars;
-		if (preset === "minimal" || preset === "full" || preset === "runner-only") {
+		if (preset === "minimal" || preset === "full") {
 			vars = getPresetVars(preset, projectRoot, tasksRootOverride);
 			console.log(`  Using preset: ${c.cyan}${preset}${c.reset}`);
 			if (tasksRootOverride) {
@@ -1955,22 +1894,6 @@ async function cmdInit(args) {
 			);
 		}
 
-		// Task runner config
-		writeFile(
-			path.join(taskplaneDir, "task-runner.yaml"),
-			generateTaskRunnerYaml(vars),
-			{ skipIfExists, label: `${configRepoName}/.taskplane/task-runner.yaml` }
-		);
-
-		// Orchestrator config (skip for runner-only preset)
-		if (preset !== "runner-only") {
-			writeFile(
-				path.join(taskplaneDir, "task-orchestrator.yaml"),
-				generateOrchestratorYaml(vars),
-				{ skipIfExists, label: `${configRepoName}/.taskplane/task-orchestrator.yaml` }
-			);
-		}
-
 		// Project config JSON (taskplane-config.json)
 		const projectConfig = generateProjectConfig(vars, initAgentConfig);
 		writeFile(
@@ -2109,10 +2032,8 @@ async function cmdInit(args) {
 		console.log(`     git push && ${c.dim}[create PR / merge to default branch]${c.reset}\n`);
 		console.log(`${c.bold}Quick start:${c.reset}`);
 		console.log(`  ${c.cyan}pi${c.reset}                                             # start pi (taskplane auto-loads)`);
-		if (preset !== "runner-only") {
-			console.log(`  ${c.cyan}/orch${c.reset}                                             # start the taskplane supervisor`);
-			console.log(`  ${c.cyan}/orch all${c.reset}                                        # run all open tasks`);
-		}
+		console.log(`  ${c.cyan}/orch${c.reset}                                             # start the taskplane supervisor`);
+		console.log(`  ${c.cyan}/orch all${c.reset}                                        # run all open tasks`);
 		if (inferTaskplaneInstallScope() === "global") {
 			console.log(`  ${c.cyan}taskplane config --save-as-defaults${c.reset}             # save these agent defaults for future inits`);
 		}
@@ -2139,7 +2060,7 @@ async function cmdInit(args) {
 
 	// Gather config values
 	let vars;
-	if (preset === "minimal" || preset === "full" || preset === "runner-only") {
+	if (preset === "minimal" || preset === "full") {
 		vars = getPresetVars(preset, projectRoot, tasksRootOverride);
 		console.log(`  Using preset: ${c.cyan}${preset}${c.reset}`);
 		if (tasksRootOverride) {
@@ -2180,23 +2101,7 @@ async function cmdInit(args) {
 		);
 	}
 
-	// Task runner config
-	writeFile(
-		path.join(projectRoot, ".pi", "task-runner.yaml"),
-		generateTaskRunnerYaml(vars),
-		{ skipIfExists, label: ".pi/task-runner.yaml" }
-	);
-
-	// Orchestrator config (skip for runner-only preset)
-	if (preset !== "runner-only") {
-		writeFile(
-			path.join(projectRoot, ".pi", "task-orchestrator.yaml"),
-			generateOrchestratorYaml(vars),
-			{ skipIfExists, label: ".pi/task-orchestrator.yaml" }
-		);
-	}
-
-	// Unified project config JSON
+	// Project config JSON
 	writeFile(
 		path.join(projectRoot, ".pi", "taskplane-config.json"),
 		JSON.stringify(generateProjectConfig(vars, initAgentConfig), null, 2) + "\n",
@@ -2271,10 +2176,8 @@ async function cmdInit(args) {
 	console.log(`\n${OK} ${c.bold}Taskplane initialized!${c.reset}\n`);
 	console.log(`${c.bold}Quick start:${c.reset}`);
 	console.log(`  ${c.cyan}pi${c.reset}                                             # start pi (taskplane auto-loads)`);
-	if (preset !== "runner-only") {
-		console.log(`  ${c.cyan}/orch${c.reset}                                             # start the taskplane supervisor`);
-		console.log(`  ${c.cyan}/orch all${c.reset}                                        # run all open tasks`);
-	}
+	console.log(`  ${c.cyan}/orch${c.reset}                                             # start the taskplane supervisor`);
+	console.log(`  ${c.cyan}/orch all${c.reset}                                        # run all open tasks`);
 	if (inferTaskplaneInstallScope() === "global") {
 		console.log(`  ${c.cyan}taskplane config --save-as-defaults${c.reset}             # save these agent defaults for future inits`);
 	}
@@ -2341,10 +2244,8 @@ function printFileList(vars, noExamples, preset, exampleTemplateDirs = [], proje
 		".pi/agents/task-reviewer.md",
 		".pi/agents/task-merger.md",
 		".pi/agents/supervisor.md",
-		".pi/task-runner.yaml",
+		".pi/taskplane-config.json",
 	];
-	if (preset !== "runner-only") files.push(".pi/task-orchestrator.yaml");
-	files.push(".pi/taskplane-config.json");
 	files.push(".pi/taskplane.json");
 	files.push(`${vars.tasks_root}/CONTEXT.md`);
 	if (!noExamples) {
@@ -2380,10 +2281,8 @@ function printWorkspaceFileList(vars, noExamples, preset, exampleTemplateDirs, c
 		`${prefix}/agents/task-reviewer.md`,
 		`${prefix}/agents/task-merger.md`,
 		`${prefix}/agents/supervisor.md`,
-		`${prefix}/task-runner.yaml`,
+		`${prefix}/taskplane-config.json`,
 	];
-	if (preset !== "runner-only") files.push(`${prefix}/task-orchestrator.yaml`);
-	files.push(`${prefix}/taskplane-config.json`);
 	files.push(`${prefix}/taskplane.json`);
 	files.push(`${prefix}/workspace.json`);
 	files.push(`${configRepoName}/${vars.tasks_root}/CONTEXT.md`);
@@ -2883,15 +2782,20 @@ function cmdDoctor() {
 	// Check project config (common — both modes)
 	console.log();
 	const hasUnifiedJson = fs.existsSync(path.join(configLocation.root, configLocation.prefix, "taskplane-config.json"));
+	const hasYamlFallback = !hasUnifiedJson && (
+		fs.existsSync(path.join(configLocation.root, configLocation.prefix, "task-runner.yaml")) ||
+		fs.existsSync(path.join(configLocation.root, configLocation.prefix, "task-orchestrator.yaml"))
+	);
 	const configFiles = [
-		{ path: "taskplane-config.json", required: false, hide: false },
-		// YAML configs are legacy fallback — hide when taskplane-config.json exists
-		{ path: "task-runner.yaml", required: !hasUnifiedJson, hide: hasUnifiedJson },
-		{ path: "task-orchestrator.yaml", required: !hasUnifiedJson, hide: hasUnifiedJson },
+		// JSON is required unless legacy YAML exists as fallback
+		{ path: "taskplane-config.json", required: !hasYamlFallback, hide: false },
+		// YAML configs are legacy fallback — only shown when JSON config is missing
+		{ path: "task-runner.yaml", required: false, hide: hasUnifiedJson },
+		{ path: "task-orchestrator.yaml", required: false, hide: hasUnifiedJson },
 		{ path: "agents/task-worker.md", required: true, hide: false },
 		{ path: "agents/task-reviewer.md", required: true, hide: false },
 		{ path: "agents/task-merger.md", required: true, hide: false },
-		// supervisor.md is created by /orch; taskplane.json is created at runtime
+		// supervisor.md is optional (scaffolded by init but may be absent in older projects); taskplane.json is created at runtime
 		{ path: "agents/supervisor.md", required: false, hide: true },
 		{ path: "taskplane.json", required: false, hide: true },
 	];
@@ -2979,7 +2883,7 @@ function cmdDoctor() {
 				console.log(`  ${OK} area '${areaName}' repo_id: ${repoId}`);
 			} else {
 				console.log(`  ${FAIL} area '${areaName}' repo_id '${repoId}' does not match any workspace repo [AREA_REPO_ID_UNKNOWN]`);
-				console.log(`     ${c.dim}→ Available repos: ${knownRepoIds.join(", ")}. Fix repo_id in ${configLocation.label}/task-runner.yaml${c.reset}`);
+				console.log(`     ${c.dim}→ Available repos: ${knownRepoIds.join(", ")}. Fix repoId in ${configLocation.label}/taskplane-config.json${c.reset}`);
 				issues++;
 			}
 		}
@@ -3239,7 +3143,7 @@ ${c.bold}Commands:${c.reset}
   ${c.cyan}help${c.reset}           Show this help message
 
 ${c.bold}Init options:${c.reset}
-  --preset <name>       Use a preset: minimal, full, runner-only
+  --preset <name>       Use a preset: minimal, full
   --tasks-root <path>   Relative tasks directory to use (e.g. docs/task-management)
   --no-examples         Skip example tasks scaffolding
   --include-examples    With --tasks-root, include example tasks (default is skip)
@@ -3261,7 +3165,7 @@ ${c.bold}Uninstall options:${c.reset}
   --package-only    Only remove installed package (skip project cleanup)
   --local           Force package uninstall from project-local scope
   --global          Force package uninstall from global scope
-  --remove-tasks    Also remove task area directories from task-runner.yaml
+  --remove-tasks    Also remove task area directories
   --all             Equivalent to --package + --remove-tasks
 
 ${c.bold}Examples:${c.reset}

package/dashboard/public/app.js

@@ -180,9 +180,9 @@ function showCopyToast(text) {
 }
 
 function copySessionId(sessionName) {
+  // Retained for potential future use but no longer rendered in the UI.
   navigator.clipboard.writeText(sessionName).then(() => {
     showCopyToast(`session ${sessionName}`);
-    // Flash the button
     const btn = document.querySelector(`[data-session="${sessionName}"]`);
     if (btn) {
       btn.classList.add("copied");
@@ -649,7 +649,7 @@ function renderLanesTasks(batch, sessions) {
     const laneSessionId = lane.laneSessionId;
     const v2Alive = isLaneAliveV2(lane.laneNumber);
     const alive = v2Alive !== null ? v2Alive : sessionSet.has(laneSessionId);
-    const sessionChip = `session: ${laneSessionId}`;
+
 
     // Lane header
     html += `<div class="lane-group">`;
@@ -670,11 +670,7 @@ function renderLanesTasks(batch, sessions) {
     // View button: shows conversation stream when available
     const isViewingConv = viewerMode === 'conversation' && viewerTarget === laneSessionId;
     html += `    <button class="session-view-btn${isViewingConv ? ' active' : ''}" onclick="viewConversation('${escapeHtml(laneSessionId)}')" title="View worker conversation">👁 View</button>`;
-    if (alive) {
-      html += `    <span class="session-cmd" data-session="${escapeHtml(laneSessionId)}" onclick="copySessionId('${escapeHtml(laneSessionId)}')" title="Copy session ID">${escapeHtml(sessionChip)}</span>`;
-    } else {
-      html += `    <span class="session-cmd dead-session">${escapeHtml(sessionChip)}</span>`;
-    }
+
     html += `  </div>`;
     html += `</div>`;
 
@@ -999,12 +995,7 @@ function renderMergeAgents(batch, sessions) {
     // Full telemetry cell
     html += `<td class="merge-telemetry-cell">${mergeTelemetryHtml(mergeTel, effectiveAlive)}</td>`;
     html += `<td>`;
-    if (effectiveAlive) {
-      const sessionChip = `session: ${effectiveSession}`;
-      html += `<span class="session-cmd" data-session="${escapeHtml(effectiveSession)}" onclick="copySessionId('${escapeHtml(effectiveSession)}')" title="Copy session ID">${escapeHtml(sessionChip)}</span>`;
-    } else {
-      html += '<span class="merge-no-data">—</span>';
-    }
+    html += '<span class="merge-no-data">—</span>';
     html += `</td>`;
     html += `<td class="merge-detail-cell">${mr.failureReason ? escapeHtml(mr.failureReason) : "—"}</td>`;
     html += `</tr>`;
@@ -1039,14 +1030,13 @@ function renderMergeAgents(batch, sessions) {
     if (shownSessions.has(sess)) continue;
 
     const sessTel = telemetry[sess] || null;
-    const sessionChip = `session: ${sess}`;
     html += `<tr>`;
     html += `<td class="merge-wave-cell">—</td>`;
     html += `<td><span class="status-badge status-running"><span class="status-dot running"></span> merging</span></td>`;
     html += `<td class="merge-session-cell">${escapeHtml(sess)}</td>`;
     // Full telemetry cell for active merge session
     html += `<td class="merge-telemetry-cell">${mergeTelemetryHtml(sessTel, true)}</td>`;
-    html += `<td><span class="session-cmd" data-session="${escapeHtml(sess)}" onclick="copySessionId('${escapeHtml(sess)}')" title="Copy session ID">${escapeHtml(sessionChip)}</span></td>`;
+    html += `<td>—</td>`;
     html += `<td>—</td>`;
     html += `</tr>`;
   }

package/extensions/taskplane/engine-worker.ts

@@ -16,6 +16,7 @@
  * @module orch/engine-worker
  */
 import type {
+	AllocatedLane,
 	EngineEvent,
 	MonitorState,
 	OrchBatchPhase,
@@ -71,6 +72,8 @@ export interface SerializedBatchState {
 	startedAt: number;
 	endedAt: number | null;
 	errors: string[];
+	/** Active lanes for the current wave (synced so /orch-sessions works). */
+	currentLanes: AllocatedLane[];
 }
 
 /**
@@ -164,6 +167,7 @@ function serializeBatchState(state: OrchBatchRuntimeState): SerializedBatchState
 		startedAt: state.startedAt,
 		endedAt: state.endedAt,
 		errors: [...state.errors],
+		currentLanes: state.currentLanes,
 	};
 }
 
@@ -192,6 +196,7 @@ export function applySerializedState(
 	batchState.startedAt = serialized.startedAt;
 	batchState.endedAt = serialized.endedAt;
 	batchState.errors = [...serialized.errors];
+	batchState.currentLanes = serialized.currentLanes ?? [];
 }
 
 // ── Engine main (runs when launched as a forked child process) ───────

package/extensions/taskplane/extension.ts

@@ -4720,7 +4720,39 @@ export default function (pi: ExtensionAPI) {
 			if (!requireExecCtx(ctx)) return;
 
 			try {
-				await openSettingsTui(ctx, execCtx!.workspaceRoot, execCtx!.pointer?.configRoot);
+				// Capture the workspace root for reload consistency — use the same
+				// root the settings TUI writes to, not ctx.cwd which may differ.
+				const reloadCwd = execCtx!.workspaceRoot;
+				await openSettingsTui(ctx, execCtx!.workspaceRoot, execCtx!.pointer?.configRoot, () => {
+					// Reload live config from disk so changes take effect immediately
+					// without requiring a session restart.
+					// Build everything into temporaries first, then commit atomically
+					// so a partial failure doesn't leave mixed-generation state.
+					try {
+						const freshCtx = buildExecutionContext(reloadCwd, loadOrchestratorConfig, loadTaskRunnerConfig);
+						let freshSupervisor: SupervisorConfig;
+						try {
+							freshSupervisor = loadSupervisorConfig(
+								freshCtx.repoRoot,
+								freshCtx.pointer?.configRoot,
+							);
+						} catch {
+							freshSupervisor = { ...DEFAULT_SUPERVISOR_CONFIG };
+						}
+						// Atomic commit — all or nothing
+						execCtx = freshCtx;
+						orchConfig = freshCtx.orchestratorConfig;
+						runnerConfig = freshCtx.taskRunnerConfig;
+						supervisorConfig = freshSupervisor;
+					} catch {
+						// Non-fatal — config was saved to disk but live reload failed.
+						// Existing in-memory config is preserved unchanged.
+						ctx.ui.notify(
+							"⚠️ Saved to disk but live reload failed. Restart to apply.",
+							"warn",
+						);
+					}
+				});
 			} catch (err: any) {
 				ctx.ui.notify(`❌ Failed to load settings: ${err.message}`, "error");
 			}

package/extensions/taskplane/settings-tui.ts

@@ -98,34 +98,36 @@ export const SECTIONS: SectionDef[] = [
 			{ configPath: "orchestrator.orchestrator.worktreeLocation", label: "Worktree Location", control: "toggle", layer: "L1", fieldType: "enum", values: ["sibling", "subdirectory"], description: "Where lane worktree directories are created" },
 			{ configPath: "orchestrator.orchestrator.worktreePrefix", label: "Worktree Prefix", control: "input", layer: "L1", fieldType: "string", description: "Prefix for worktree directory names" },
 			{ configPath: "orchestrator.orchestrator.batchIdFormat", label: "Batch ID Format", control: "toggle", layer: "L1", fieldType: "enum", values: ["timestamp", "sequential"], description: "Batch ID format for logs/branch naming" },
-			// spawn_mode removed from Orchestrator section — Runtime V2 is subprocess-only.
-			// The user-facing spawn mode setting is under Worker (controls /task behavior).
 			{ configPath: "orchestrator.orchestrator.sessionPrefix", label: "Session Prefix", control: "input", layer: "L1+L2", fieldType: "string", prefsKey: "sessionPrefix", description: "Prefix for orchestrator session names" },
 			{ configPath: "orchestrator.orchestrator.operatorId", label: "Operator ID", control: "input", layer: "L1+L2", fieldType: "string", prefsKey: "operatorId", description: "Operator identifier (empty = auto-detect)" },
 			{ configPath: "orchestrator.orchestrator.integration", label: "Integration", control: "picker", layer: "L1", fieldType: "enum", values: ["manual", "supervised", "auto"], description: "How completed batches are integrated. manual = user runs /orch-integrate. supervised = supervisor proposes plan, asks confirmation. auto = supervisor executes without asking." },
 		],
 	},
 	{
-		name: "Dependencies",
+		name: "Agent: Supervisor",
 		fields: [
-			{ configPath: "orchestrator.dependencies.source", label: "Dep Source", control: "toggle", layer: "L1", fieldType: "enum", values: ["prompt", "agent"], description: "Dependency extraction source" },
-			{ configPath: "orchestrator.dependencies.cache", label: "Dep Cache", control: "toggle", layer: "L1", fieldType: "boolean", values: ["true", "false"], description: "Cache dependency analysis results" },
+			{ configPath: "orchestrator.supervisor.model", label: "Supervisor Model", control: "input", layer: "L1+L2", fieldType: "string", prefsKey: "supervisorModel", description: "Supervisor model (inherit = use session model)" },
+			{ configPath: "orchestrator.supervisor.autonomy", label: "Autonomy Level", control: "picker", layer: "L1", fieldType: "enum", values: ["interactive", "supervised", "autonomous"], description: "Recovery action confirmation behavior" },
 		],
 	},
 	{
-		name: "Assignment",
+		name: "Agent: Worker",
 		fields: [
-			{ configPath: "orchestrator.assignment.strategy", label: "Strategy", control: "toggle", layer: "L1", fieldType: "enum", values: ["affinity-first", "round-robin", "load-balanced"], description: "Lane assignment strategy" },
+			{ configPath: "taskRunner.worker.model", label: "Worker Model", control: "input", layer: "L1+L2", fieldType: "string", prefsKey: "workerModel", description: "Worker model (inherit = use session model)" },
+			{ configPath: "taskRunner.worker.tools", label: "Worker Tools", control: "input", layer: "L1", fieldType: "string", description: "Worker tool allowlist" },
+			{ configPath: "taskRunner.worker.thinking", label: "Worker Thinking", control: "picker", layer: "L1", fieldType: "string", description: "Worker thinking mode" },
 		],
 	},
 	{
-		name: "Pre-Warm",
+		name: "Agent: Reviewer",
 		fields: [
-			{ configPath: "orchestrator.preWarm.autoDetect", label: "Auto-Detect", control: "toggle", layer: "L1", fieldType: "boolean", values: ["true", "false"], description: "Enable automatic pre-warm command detection" },
+			{ configPath: "taskRunner.reviewer.model", label: "Reviewer Model", control: "input", layer: "L1+L2", fieldType: "string", prefsKey: "reviewerModel", description: "Reviewer model (inherit = use session model)" },
+			{ configPath: "taskRunner.reviewer.tools", label: "Reviewer Tools", control: "input", layer: "L1", fieldType: "string", description: "Reviewer tool allowlist" },
+			{ configPath: "taskRunner.reviewer.thinking", label: "Reviewer Thinking", control: "picker", layer: "L1", fieldType: "string", description: "Reviewer thinking mode" },
 		],
 	},
 	{
-		name: "Merge",
+		name: "Agent: Merge",
 		fields: [
 			{ configPath: "orchestrator.merge.model", label: "Merge Model", control: "input", layer: "L1+L2", fieldType: "string", prefsKey: "mergeModel", description: "Merge-agent model (inherit = use session model)" },
 			{ configPath: "orchestrator.merge.tools", label: "Merge Tools", control: "input", layer: "L1", fieldType: "string", description: "Merge-agent tool allowlist" },
@@ -134,6 +136,18 @@ export const SECTIONS: SectionDef[] = [
 			{ configPath: "orchestrator.merge.timeoutMinutes", label: "Merge Timeout (minutes)", control: "input", layer: "L1", fieldType: "number", description: "Max time for merge agent to complete. Increase for large batches (default: 10)" },
 		],
 	},
+	{
+		name: "Context Limits",
+		fields: [
+			{ configPath: "taskRunner.context.workerContextWindow", label: "Context Window", control: "input", layer: "L1", fieldType: "number", description: "Worker context window size" },
+			{ configPath: "taskRunner.context.warnPercent", label: "Warn %", control: "input", layer: "L1", fieldType: "number", description: "Context utilization warn threshold (%)" },
+			{ configPath: "taskRunner.context.killPercent", label: "Kill %", control: "input", layer: "L1", fieldType: "number", description: "Context utilization hard-stop threshold (%)" },
+			{ configPath: "taskRunner.context.maxWorkerIterations", label: "Max Iterations", control: "input", layer: "L1", fieldType: "number", description: "Max worker iterations per step" },
+			{ configPath: "taskRunner.context.maxReviewCycles", label: "Max Review Cycles", control: "input", layer: "L1", fieldType: "number", description: "Max revise loops per review stage" },
+			{ configPath: "taskRunner.context.noProgressLimit", label: "No Progress Limit", control: "input", layer: "L1", fieldType: "number", description: "Max no-progress iterations before failure" },
+			{ configPath: "taskRunner.context.maxWorkerMinutes", label: "Max Worker Min (ctx)", control: "input", layer: "L1", fieldType: "number", optional: true, description: "Per-worker wall-clock cap (minutes, empty = no cap)" },
+		],
+	},
 	{
 		name: "Failure Policy",
 		fields: [
@@ -145,46 +159,28 @@ export const SECTIONS: SectionDef[] = [
 		],
 	},
 	{
-		name: "Monitoring",
-		fields: [
-			{ configPath: "orchestrator.monitoring.pollInterval", label: "Poll Interval (sec)", control: "input", layer: "L1", fieldType: "number", description: "Poll interval for lane/task monitoring (seconds)" },
-		],
-	},
-	{
-		name: "Supervisor",
+		name: "Dependencies",
 		fields: [
-			{ configPath: "orchestrator.supervisor.model", label: "Supervisor Model", control: "input", layer: "L1+L2", fieldType: "string", prefsKey: "supervisorModel", description: "Supervisor model (inherit = use session model)" },
-			{ configPath: "orchestrator.supervisor.autonomy", label: "Autonomy Level", control: "picker", layer: "L1", fieldType: "enum", values: ["interactive", "supervised", "autonomous"], description: "Recovery action confirmation behavior" },
+			{ configPath: "orchestrator.dependencies.source", label: "Dep Source", control: "toggle", layer: "L1", fieldType: "enum", values: ["prompt", "agent"], description: "Dependency extraction source" },
+			{ configPath: "orchestrator.dependencies.cache", label: "Dep Cache", control: "toggle", layer: "L1", fieldType: "boolean", values: ["true", "false"], description: "Cache dependency analysis results" },
 		],
 	},
 	{
-		name: "Worker",
+		name: "Assignment",
 		fields: [
-			{ configPath: "taskRunner.worker.model", label: "Worker Model", control: "input", layer: "L1+L2", fieldType: "string", prefsKey: "workerModel", description: "Worker model (inherit = use session model)" },
-			{ configPath: "taskRunner.worker.tools", label: "Worker Tools", control: "input", layer: "L1", fieldType: "string", description: "Worker tool allowlist" },
-			{ configPath: "taskRunner.worker.thinking", label: "Worker Thinking", control: "picker", layer: "L1", fieldType: "string", description: "Worker thinking mode" },
-			// spawnMode removed — /task is deprecated, Runtime V2 is subprocess-only.
-			// { configPath: "taskRunner.worker.spawnMode" ... } was here.
+			{ configPath: "orchestrator.assignment.strategy", label: "Strategy", control: "toggle", layer: "L1", fieldType: "enum", values: ["affinity-first", "round-robin", "load-balanced"], description: "Lane assignment strategy" },
 		],
 	},
 	{
-		name: "Reviewer",
+		name: "Pre-Warm",
 		fields: [
-			{ configPath: "taskRunner.reviewer.model", label: "Reviewer Model", control: "input", layer: "L1+L2", fieldType: "string", prefsKey: "reviewerModel", description: "Reviewer model (inherit = use session model)" },
-			{ configPath: "taskRunner.reviewer.tools", label: "Reviewer Tools", control: "input", layer: "L1", fieldType: "string", description: "Reviewer tool allowlist" },
-			{ configPath: "taskRunner.reviewer.thinking", label: "Reviewer Thinking", control: "picker", layer: "L1", fieldType: "string", description: "Reviewer thinking mode" },
+			{ configPath: "orchestrator.preWarm.autoDetect", label: "Auto-Detect", control: "toggle", layer: "L1", fieldType: "boolean", values: ["true", "false"], description: "Enable automatic pre-warm command detection" },
 		],
 	},
 	{
-		name: "Context Limits",
+		name: "Monitoring",
 		fields: [
-			{ configPath: "taskRunner.context.workerContextWindow", label: "Context Window", control: "input", layer: "L1", fieldType: "number", description: "Worker context window size" },
-			{ configPath: "taskRunner.context.warnPercent", label: "Warn %", control: "input", layer: "L1", fieldType: "number", description: "Context utilization warn threshold (%)" },
-			{ configPath: "taskRunner.context.killPercent", label: "Kill %", control: "input", layer: "L1", fieldType: "number", description: "Context utilization hard-stop threshold (%)" },
-			{ configPath: "taskRunner.context.maxWorkerIterations", label: "Max Iterations", control: "input", layer: "L1", fieldType: "number", description: "Max worker iterations per step" },
-			{ configPath: "taskRunner.context.maxReviewCycles", label: "Max Review Cycles", control: "input", layer: "L1", fieldType: "number", description: "Max revise loops per review stage" },
-			{ configPath: "taskRunner.context.noProgressLimit", label: "No Progress Limit", control: "input", layer: "L1", fieldType: "number", description: "Max no-progress iterations before failure" },
-			{ configPath: "taskRunner.context.maxWorkerMinutes", label: "Max Worker Min (ctx)", control: "input", layer: "L1", fieldType: "number", optional: true, description: "Per-worker wall-clock cap (minutes, empty = no cap)" },
+			{ configPath: "orchestrator.monitoring.pollInterval", label: "Poll Interval (sec)", control: "input", layer: "L1", fieldType: "number", description: "Poll interval for lane/task monitoring (seconds)" },
 		],
 	},
 	{
@@ -1166,9 +1162,10 @@ export async function openSettingsTui(
 	ctx: ExtensionContext,
 	configRoot: string,
 	pointerConfigRoot?: string,
+	onConfigChanged?: () => void,
 ): Promise<void> {
 	// Load current config state — refreshed each time we return to the top level
-	await showSectionSelectorLoop(ctx, configRoot, pointerConfigRoot);
+	await showSectionSelectorLoop(ctx, configRoot, pointerConfigRoot, onConfigChanged);
 }
 
 /**
@@ -1200,6 +1197,7 @@ async function showSectionSelectorLoop(
 	ctx: ExtensionContext,
 	configRoot: string,
 	pointerConfigRoot?: string,
+	onConfigChanged?: () => void,
 ): Promise<void> {
 	while (true) {
 		const state = loadConfigState(configRoot, pointerConfigRoot);
@@ -1257,7 +1255,7 @@ async function showSectionSelectorLoop(
 		if (section.readOnly) {
 			await showAdvancedSection(ctx, state.mergedConfig);
 		} else {
-			await showSectionSettingsLoop(ctx, section, configRoot, pointerConfigRoot);
+			await showSectionSettingsLoop(ctx, section, configRoot, pointerConfigRoot, onConfigChanged);
 		}
 	}
 }
@@ -1339,6 +1337,7 @@ async function showSectionSettingsLoop(
 	section: SectionDef,
 	configRoot: string,
 	pointerConfigRoot?: string,
+	onConfigChanged?: () => void,
 ): Promise<void> {
 	while (true) {
 		const state = loadConfigState(configRoot, pointerConfigRoot);
@@ -1438,9 +1437,13 @@ async function showSectionSettingsLoop(
 			} else {
 				writeGlobalPreference(toGlobalPreferencePath(field), typedValue);
 			}
+			// Notify caller to reload in-memory config from disk
+			if (onConfigChanged) {
+				try { onConfigChanged(); } catch { /* non-fatal */ }
+			}
+
 			ctx.ui.notify(
-				`✅ ${field.label} updated.\n` +
-				`ℹ Restart session to apply changes.`,
+				`✅ ${field.label} updated.`,
 				"info",
 			);
 
@@ -1494,9 +1497,38 @@ async function showSectionSettingsOnce(
 			description: field.description,
 		};
 
-		// Toggle fields get values array for cycling
+		// Toggle fields: use cycling for 2 values (boolean-like), submenu for 3+
 		if (field.control === "toggle" && field.values) {
-			item.values = field.values.map((v) => `${v}  ${sourceBadge}`);
+			if (field.values.length <= 2) {
+				item.values = field.values.map((v) => `${v}  ${sourceBadge}`);
+			} else {
+				// 3+ values: use a submenu so the user can pick any option,
+				// not just cycle to the next one and immediately commit.
+				item.submenu = (_currentValue: string, done: (selected?: string) => void) => {
+					const selectItems: SelectItem[] = field.values!.map((v) => ({
+						value: `${v}  ${sourceBadge}`,
+						label: v,
+					}));
+					// Return SelectList directly — Container doesn't forward
+					// handleInput to children, which would freeze the TUI.
+					const list = new SelectList(
+						selectItems,
+						Math.min(selectItems.length + 1, 10),
+						{
+							selectedPrefix: (t: string) => `\x1b[36m${t}\x1b[0m`,
+							selectedText: (t: string) => `\x1b[36m${t}\x1b[0m`,
+							description: (t: string) => `\x1b[2m${t}\x1b[0m`,
+							scrollInfo: (t: string) => `\x1b[2m${t}\x1b[0m`,
+							noMatch: (t: string) => `\x1b[33m${t}\x1b[0m`,
+						},
+					);
+					const currentIdx = field.values!.indexOf(displayValue);
+					if (currentIdx >= 0) list.setSelectedIndex(currentIdx);
+					list.onSelect = (selected) => done(selected.value);
+					list.onCancel = () => done();
+					return list;
+				};
+			}
 		}
 
 		// Input fields: use a single-value cycling pattern instead of a submenu.
@@ -1649,15 +1681,16 @@ function truncateLine(text: string, width: number): string {
  */
 const SECTION_CONFIG_PREFIXES: Record<string, string[]> = {
 	"Orchestrator": ["orchestrator.orchestrator"],
+	"Agent: Supervisor": ["orchestrator.supervisor"],
+	"Agent: Worker": ["taskRunner.worker"],
+	"Agent: Reviewer": ["taskRunner.reviewer"],
+	"Agent: Merge": ["orchestrator.merge"],
+	"Context Limits": ["taskRunner.context"],
+	"Failure Policy": ["orchestrator.failure"],
 	"Dependencies": ["orchestrator.dependencies"],
 	"Assignment": ["orchestrator.assignment"],
 	"Pre-Warm": ["orchestrator.preWarm"],
-	"Merge": ["orchestrator.merge"],
-	"Failure Policy": ["orchestrator.failure"],
 	"Monitoring": ["orchestrator.monitoring"],
-	"Worker": ["taskRunner.worker"],
-	"Reviewer": ["taskRunner.reviewer"],
-	"Context Limits": ["taskRunner.context"],
 };
 
 /**

package/extensions/taskplane/supervisor.ts

@@ -951,7 +951,7 @@ export function triggerSupervisorIntegration(
 				}],
 				display: "Integration plan ready — awaiting operator confirmation",
 			},
-			{ triggerTurn: true, deliverAs: "nextTurn" },
+			{ triggerTurn: true },
 		);
 
 		// TP-043 R004: Defer summary until after integration completes (or operator declines).
@@ -2834,7 +2834,11 @@ export async function activateSupervisor(
 				],
 				display: `Supervisor activated — ${routingContext.routingState}`,
 			},
-			{ triggerTurn: true, deliverAs: "nextTurn" },
+			// triggerTurn starts an LLM turn immediately when the agent is idle.
+			// Do NOT use deliverAs:"nextTurn" — that queues the message for a
+			// future turn instead of starting one, causing the terminal to hang
+			// until the user sends input.
+			{ triggerTurn: true },
 		);
 		return;
 	}
@@ -2886,7 +2890,9 @@ export async function activateSupervisor(
 			],
 			display: "Supervisor activated" + (batchState.batchId ? ` for batch ${batchState.batchId}` : ""),
 		},
-		{ triggerTurn: true, deliverAs: "nextTurn" },
+		// triggerTurn starts an LLM turn immediately when the agent is idle.
+		// Do NOT use deliverAs:"nextTurn" here — see routing path comment.
+		{ triggerTurn: true },
 	);
 }
 
@@ -3032,7 +3038,7 @@ export async function transitionToRoutingMode(
 			}],
 			display: `Supervisor — ${routingContext.routingState}`,
 		},
-		{ triggerTurn: true, deliverAs: "nextTurn" },
+		{ triggerTurn: true },
 	);
 }
 
@@ -4241,7 +4247,7 @@ export function startEventTailer(
 				content: [{ type: "text", text }],
 				display: text.replace(/\*\*/g, "").substring(0, 80),
 			},
-			{ triggerTurn: true, deliverAs: "nextTurn" },
+			{ triggerTurn: true },
 		);
 	};
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "taskplane",
-  "version": "0.24.31",
+  "version": "0.25.0",
   "description": "AI agent orchestration for pi — parallel task execution with checkpoint discipline",
   "keywords": [
     "pi-package",

package/templates/config/task-orchestrator.yaml

@@ -1,93 +0,0 @@
-# ═══════════════════════════════════════════════════════════════════════
-# Parallel Task Orchestrator Configuration
-# ═══════════════════════════════════════════════════════════════════════
-#
-# Copy this file to `.pi/task-orchestrator.yaml` in your project and
-# customize it. The orchestrator reads BOTH this file and task-runner.yaml.
-#
-#   - task-runner.yaml        → task areas, reference docs, worker/reviewer
-#   - task-orchestrator.yaml  → lane count, worktrees, merge, failure policy
-#
-# This template is intentionally conservative so it is safe to adapt to a
-# wide range of repositories.
-# ═══════════════════════════════════════════════════════════════════════
-
-# ── Orchestrator Core ─────────────────────────────────────────────────
-
-orchestrator:
-  # Maximum parallel lanes (worktrees)
-  max_lanes: 3
-
-  # Where to create worktree directories.
-  # "sibling"       = ../{prefix}-{opId}-{N}          (e.g. ../project-wt-alice-1)
-  # "subdirectory"  = .worktrees/{prefix}-{opId}-{N}  (e.g. .worktrees/project-wt-alice-1)
-  worktree_location: "subdirectory"
-  worktree_prefix: "project-wt"
-
-  # Batch ID format used in branch names and logs.
-  batch_id_format: "timestamp"
-
-  # Runtime V2 execution backend (subprocess-only)
-  spawn_mode: "subprocess"
-
-  # Prefix for orchestrator session names.
-  session_prefix: "orch"
-
-  # Optional operator identifier for team-scale collision resistance.
-  # Auto-detected from OS username if empty. Set explicitly in CI or
-  # when multiple operators share the same machine.
-  # operator_id: ""
-
-# ── Dependency Analysis ───────────────────────────────────────────────
-
-dependencies:
-  # "prompt" = parse dependencies from PROMPT.md
-  # "agent"  = use an agent to analyze tasks
-  source: "prompt"
-  cache: true
-
-# ── Lane Assignment ───────────────────────────────────────────────────
-
-assignment:
-  strategy: "affinity-first"
-  size_weights:
-    S: 1
-    M: 2
-    L: 4
-
-# ── Pre-warming ───────────────────────────────────────────────────────
-
-# Disabled by default. Add commands that fit your stack if you want to use it.
-pre_warm:
-  auto_detect: false
-  commands: {}
-  always: []
-
-# ── Merge ─────────────────────────────────────────────────────────────
-
-merge:
-  model: ""                    # empty = inherit from parent pi session
-  tools: "read,write,edit,bash,grep,find,ls"
-
-  # Verification commands to run after each merge.
-  # Add only the commands that are safe and relevant for your project.
-  verify: []
-
-  order: "fewest-files-first"
-
-  # Merge agent timeout in minutes. Increase for large batches with many files.
-  timeout_minutes: 90
-
-# ── Failure Handling ──────────────────────────────────────────────────
-
-failure:
-  on_task_failure: "skip-dependents"
-  on_merge_failure: "pause"
-  stall_timeout: 30
-  max_worker_minutes: 30
-  abort_grace_period: 60
-
-# ── Monitoring ────────────────────────────────────────────────────────
-
-monitoring:
-  poll_interval: 5

package/templates/config/task-runner.yaml

@@ -1,94 +0,0 @@
-# ═══════════════════════════════════════════════════════════════════════
-# Task Runner Configuration
-# ═══════════════════════════════════════════════════════════════════════
-#
-# Copy this file to `.pi/task-runner.yaml` in your project and customize it.
-# This template is intentionally generic — replace the example paths, test
-# commands, and task areas with ones that fit your repository.
-# ═══════════════════════════════════════════════════════════════════════
-
-# ── Project ───────────────────────────────────────────────────────────
-
-project:
-  name: "Your Project"
-  description: "Short description of your project"
-
-paths:
-  tasks: "tasks"
-  architecture: "docs/architecture.md"
-
-# ── Verification Commands ─────────────────────────────────────────────
-
-# Add the commands your project uses for validation. Keep only the ones
-# that are relevant to your stack.
-testing:
-  commands:
-    test: "npm test"
-    build: "npm run build"
-    lint: "npm run lint"
-
-# ── Standards ─────────────────────────────────────────────────────────
-
-standards:
-  docs:
-    - "README.md"
-    - "CONTRIBUTING.md"
-  rules:
-    - "Keep changes scoped to the task"
-    - "Update documentation when behavior changes"
-    - "Prefer typed interfaces over unstructured data"
-    - "Avoid destructive changes unless explicitly requested"
-
-# Per-area standards overrides. Omit or delete if you do not need them.
-standards_overrides: {}
-
-# ── Runner Settings ───────────────────────────────────────────────────
-
-worker:
-  model: ""                    # empty = inherit from parent pi session
-  tools: "read,write,edit,bash,grep,find,ls"
-  thinking: ""                 # empty = inherit from parent pi session
-  # spawn_mode: "subprocess"   # currently supported runtime mode
-
-reviewer:
-  model: ""
-  tools: "read,write,bash,grep,find,ls"
-  thinking: "off"
-
-context:
-  # worker_context_window: 200000  # 0 or omit = auto-detect from model registry; set explicitly to override
-  warn_percent: 85
-  kill_percent: 95
-  max_worker_iterations: 20
-  max_review_cycles: 2
-  no_progress_limit: 3
-  # max_worker_minutes: 30      # optional wall-clock guard for long worker runs
-
-# ── Task Creation / Discovery ─────────────────────────────────────────
-
-# Define the task areas that exist in your project.
-task_areas:
-  general:
-    path: "taskplane-tasks"
-    prefix: "TP"
-    context: "taskplane-tasks/CONTEXT.md"
-
-# Reference docs available for higher-context task prompts.
-reference_docs:
-  overview: "README.md"
-  architecture: "docs/architecture.md"
-  contributing: "CONTRIBUTING.md"
-
-# Docs that should never be loaded during task execution.
-never_load:
-  - "PROGRESS.md"
-  - "HANDOFF-LOG.md"
-
-# Self-documentation targets (where agents should log useful discoveries).
-self_doc_targets:
-  tech_debt: "CONTEXT.md ## Technical Debt / Future Work"
-
-# Docs requiring explicit user approval to modify.
-protected_docs:
-  - "docs/"
-  - "templates/"