nrslib · nrslib · Feb 14, 2026 · Feb 13, 2026 · Feb 13, 2026 · Feb 13, 2026
diff --git a/.gitignore b/.gitignore
@@ -33,3 +33,7 @@ coverage/
 
 task_planning/
 
+OPENCODE_CONFIG_CONTENT
+
+# Local editor/agent settings
+.claude/
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -4,6 +4,37 @@ All notable changes to this project will be documented in this file.
 
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/).
 
+## [0.14.0] - 2026-02-14
+
+### Added
+
+- **`takt list` インストラクトモード (#267)**: 既存ブランチに対して追加指示を行えるインストラクトモードを追加 — 会話ループで要件を詳細化してからピース実行が可能に
+- **`takt list` 完了タスクアクション (#271)**: 完了タスクに対する diff 表示・ブランチ操作（マージ、削除）を追加
+- **Claude サンドボックス設定**: `provider_options.claude.sandbox` でサンドボックスの除外コマンド（`excluded_commands`）やサンドボックス無効化（`allow_unsandboxed_commands`）を設定可能に
+- **`provider_options` のグローバル/プロジェクト設定**: `provider_options` を `~/.takt/config.yaml`（グローバル）および `.takt/config.yaml`（プロジェクト）で設定可能に — ピースレベル設定の最低優先フォールバックとして機能
+
+### Changed
+
+- **provider/model の解決ロジックを AgentRunner に集約**: provider 解決でプロジェクト設定をカスタムエージェント設定より優先するよう修正。ステップレベルの `stepModel` / `stepProvider` による上書きを追加
+- **ポストエクスキューションの共通化**: インタラクティブモードとインストラクトモードで post-execution フロー（auto-commit, push, PR 作成）を `postExecution.ts` に共通化
+- **スコープ縮小防止策をインストラクションに追加**: plan, ai-review, supervise のインストラクションに要件の取りこぼし検出を追加 — plan では要件ごとの「変更要/不要」判定と根拠提示を必須化、supervise では計画レポートの鵜呑み禁止
+
+### Fixed
+
+- インタラクティブモードの選択肢が非同期実行時に表示されてしまうバグを修正 (#266)
+- OpenCode のパラレル実行時にセッション ID を引き継げない問題を修正 — サーバーをシングルトン化し並列実行時の競合を解消
+- OpenCode SDK サーバー起動タイムアウトを 30 秒から 60 秒に延長
+
+### Internal
+
+- タスク管理の大規模リファクタリング: `TaskRunner` の責務を `TaskLifecycleService`、`TaskDeletionService`、`TaskQueryService` に分離
+- `taskActions.ts` を機能別に分割: `taskBranchLifecycleActions.ts`、`taskDiffActions.ts`、`taskInstructionActions.ts`、`taskDeleteActions.ts`
+- `postExecution.ts`、`taskResultHandler.ts`、`instructMode.ts`、`taskActionTarget.ts` を新規追加
+- ピース選択ロジックを `pieceSelection/index.ts` に集約（`selectAndExecute.ts` から抽出）
+- テスト追加: instructMode, listNonInteractive-completedActions, listTasksInteractiveStatusActions, option-resolution-order, taskInstructionActions, selectAndExecute-autoPr 等を新規・拡充
+- E2E テストに Claude Code サンドボックス対応オプション（`dangerouslyDisableSandbox`）を追加
+- `OPENCODE_CONFIG_CONTENT` を `.gitignore` に追加
+
 ## [0.13.0] - 2026-02-13
 
 ### Added

diff --git a/builtins/en/instructions/ai-review.md b/builtins/en/instructions/ai-review.md
@@ -8,6 +8,7 @@ Review the code for AI-specific issues:
 - Plausible but incorrect patterns
 - Compatibility with the existing codebase
 - Scope creep detection
+- Scope shrinkage detection (missing task requirements)
 
 ## Judgment Procedure
 

diff --git a/builtins/en/instructions/plan.md b/builtins/en/instructions/plan.md
@@ -12,6 +12,7 @@ For small tasks, skip the design sections in the report.
 
 **Actions:**
 1. Understand the task requirements
+   - **For each requirement, determine "change needed / not needed". If "not needed", cite the relevant code (file:line) as evidence. Claiming "already correct" without evidence is prohibited**
 2. Investigate code to resolve unknowns
 3. Identify the impact area
 4. Determine file structure and design patterns (if needed)

diff --git a/builtins/en/instructions/supervise.md b/builtins/en/instructions/supervise.md
@@ -3,7 +3,8 @@ Run tests, verify the build, and perform final approval.
 **Overall piece verification:**
 1. Whether the plan and implementation results are consistent
 2. Whether findings from each review movement have been addressed
-3. Whether the original task objective has been achieved
+3. Whether each task spec requirement has been achieved
+   - Do not rely on the plan report's judgment; independently verify each requirement against actual code (file:line)
 
 **Report verification:** Read all reports in the Report Directory and
 check for any unaddressed improvement suggestions.

diff --git a/builtins/ja/instructions/ai-review.md b/builtins/ja/instructions/ai-review.md
@@ -8,6 +8,7 @@ AI特有の問題についてコードをレビューしてください:
 - もっともらしいが間違っているパターン
 - 既存コードベースとの適合性
 - スコープクリープの検出
+- スコープ縮小の検出（タスク要件の取りこぼし）
 
 ## 判定手順
 

diff --git a/builtins/ja/instructions/plan.md b/builtins/ja/instructions/plan.md
@@ -18,6 +18,7 @@
    - **指示書に明記されていない別ファイルを「参照資料の代わり」として使うことは禁止**
 2. タスクの要件を理解する
    - 参照資料の内容と現在の実装を突き合わせて差分を特定する
+   - **要件ごとに「変更要/不要」を判定する。「不要」の場合は現行コードの該当箇所（ファイル:行）を根拠として示すこと。根拠なしの「既に正しい」は禁止**
 3. コードを調査して不明点を解決する
 4. 影響範囲を特定する
 5. ファイル構成・設計パターンを決定する（必要な場合）

diff --git a/builtins/ja/instructions/supervise.md b/builtins/ja/instructions/supervise.md
@@ -3,7 +3,8 @@
 **ピース全体の確認:**
 1. 計画と実装結果が一致しているか
 2. 各レビュームーブメントの指摘が対応されているか
-3. 元のタスク目的が達成されているか
+3. タスク指示書の各要件が達成されているか
+   - 計画レポートの判断を鵜呑みにせず、要件ごとに実コード（ファイル:行）で独立照合する
 
 **レポートの確認:** Report Directory内の全レポートを読み、
 未対応の改善提案がないか確認してください。

diff --git a/docs/provider-sandbox.md b/docs/provider-sandbox.md
@@ -0,0 +1,168 @@
+# Provider Sandbox Configuration
+
+TAKT supports configuring sandbox settings for AI agent providers. This document covers how sandbox isolation works across providers, how to configure it, and the security trade-offs.
+
+## Overview
+
+| Provider | Sandbox Mechanism | Build Tool Issues | TAKT Configuration |
+|----------|------------------|-------------------|-------------------|
+| **Claude Code** | macOS Seatbelt / Linux bubblewrap | Gradle/JVM blocked in `edit` mode | `provider_options.claude.sandbox` |
+| **Codex CLI** | macOS Seatbelt / Linux Landlock+seccomp | npm/maven/pytest failures (widespread) | `provider_options.codex.network_access` |
+| **OpenCode CLI** | None (no native sandbox) | No constraints (no security either) | N/A |
+
+## Claude Code Sandbox
+
+### The Problem
+
+When a movement uses `permission_mode: edit` (mapped to Claude SDK's `acceptEdits`), Bash commands run inside a macOS Seatbelt sandbox. This sandbox blocks:
+
+- Writes outside the working directory (e.g., `~/.gradle`)
+- Certain system calls required by JVM initialization
+- Network access (by default)
+
+As a result, build tools like Gradle, Maven, or any JVM-based tool fail with `Operation not permitted`.
+
+### Solution: `provider_options.claude.sandbox`
+
+TAKT exposes Claude SDK's `SandboxSettings` through `provider_options.claude.sandbox` at four configuration levels.
+
+#### Option A: `allow_unsandboxed_commands` (Recommended)
+
+Allow all Bash commands to run outside the sandbox while keeping file edit permissions controlled:
+
+```yaml
+provider_options:
+  claude:
+    sandbox:
+      allow_unsandboxed_commands: true
+```
+
+#### Option B: `excluded_commands`
+
+Exclude only specific commands from the sandbox:
+
+```yaml
+provider_options:
+  claude:
+    sandbox:
+      excluded_commands:
+        - ./gradlew
+        - npm
+        - npx
+```
+
+### Configuration Levels
+
+Settings are merged with the following priority (highest wins):
+
+```
+Movement > Piece > Project Local > Global
+```
+
+#### Global (`~/.takt/config.yaml`)
+
+Applies to all projects and all pieces:
+
+```yaml
+# ~/.takt/config.yaml
+provider_options:
+  claude:
+    sandbox:
+      allow_unsandboxed_commands: true
+```
+
+#### Project Local (`.takt/config.yaml`)
+
+Applies to this project only:
+
+```yaml
+# .takt/config.yaml
+provider_options:
+  claude:
+    sandbox:
+      excluded_commands:
+        - ./gradlew
+```
+
+#### Piece (`piece_config` section)
+
+Applies to all movements in this piece:
+
+```yaml
+# pieces/my-piece.yaml
+piece_config:
+  provider_options:
+    claude:
+      sandbox:
+        allow_unsandboxed_commands: true
+```
+
+#### Movement (per step)
+
+Applies to a specific movement only:
+
+```yaml
+movements:
+  - name: implement
+    permission_mode: edit
+    provider_options:
+      claude:
+        sandbox:
+          allow_unsandboxed_commands: true
+  - name: review
+    permission_mode: readonly
+    # No sandbox config needed — readonly doesn't sandbox Bash
+```
+
+### Security Risk Comparison
+
+| Configuration | File Edits | Network | Bash Commands | CWD-external Writes | Risk Level |
+|--------------|-----------|---------|---------------|---------------------|------------|
+| `permission_mode: edit` (default) | Permitted | Blocked | Sandboxed | Blocked | Low |
+| `excluded_commands: [./gradlew]` | Permitted | Blocked | Only `./gradlew` unsandboxed | Only via `./gradlew` | Low |
+| `allow_unsandboxed_commands: true` | Permitted | Allowed | Unsandboxed | Allowed via Bash | **Medium** |
+| `permission_mode: full` | All permitted | Allowed | Unsandboxed | All permitted | **High** |
+
+**Key difference between `allow_unsandboxed_commands` and `permission_mode: full`:**
+- `allow_unsandboxed_commands`: File edits still require Claude Code's permission check (`acceptEdits` mode). Only Bash is unsandboxed.
+- `permission_mode: full`: All permission checks are bypassed (`bypassPermissions` mode). No guardrails at all.
+
+### Practical Risk Assessment
+
+The "Medium" risk of `allow_unsandboxed_commands` is manageable in practice because:
+
+- TAKT runs locally on the developer's machine (not a public-facing service)
+- Input comes from task instructions written by the developer
+- Agent behavior is reviewed by the supervisor movement
+- File edit operations still go through Claude Code's permission system
+
+## Codex CLI Sandbox
+
+Codex CLI uses macOS Seatbelt (same as Claude Code) but has **more severe compatibility issues** with build tools. Community reports show npm, Maven, pytest, and other tools frequently failing with `Operation not permitted` — even when the same commands work in Claude Code.
+
+Codex sandbox is configured via `~/.codex/config.toml` (not through TAKT):
+
+```toml
+# ~/.codex/config.toml
+sandbox_mode = "workspace-write"
+
+[sandbox_workspace_write]
+network_access = true
+writable_roots = ["/Users/YOU/.gradle"]
+```
+
+TAKT provides `provider_options.codex.network_access` to control network access via the Codex SDK:
+
+```yaml
+provider_options:
+  codex:
+    network_access: true
+```
+
+For other sandbox settings (writable_roots, sandbox_mode), configure directly in `~/.codex/config.toml`.
+
+## OpenCode CLI Sandbox
+
+OpenCode CLI does not have a native sandbox mechanism. All commands run without filesystem or network restrictions. For isolation, the community recommends Docker containers (e.g., [opencode-sandbox](https://github.com/fabianlema/opencode-sandbox)).
+
+No TAKT-side sandbox configuration is needed or available for OpenCode.
diff --git a/e2e/specs/run-sigint-graceful.e2e.ts b/e2e/specs/run-sigint-graceful.e2e.ts
@@ -221,7 +221,6 @@ describe('E2E: Run tasks graceful shutdown on SIGINT (parallel)', () => {
       env: {
         ...isolatedEnv.env,
         TAKT_MOCK_SCENARIO: scenarioPath,
-        TAKT_E2E_SELF_SIGINT_TWICE: '1',
       },
       stdio: ['ignore', 'pipe', 'pipe'],
     });
@@ -242,9 +241,14 @@ describe('E2E: Run tasks graceful shutdown on SIGINT (parallel)', () => {
     );
     expect(workersFilled, `stdout:\n${stdout}\n\nstderr:\n${stderr}`).toBe(true);
 
+    // Simulate user pressing Ctrl+C twice.
+    child.kill('SIGINT');
+    await new Promise((resolvePromise) => setTimeout(resolvePromise, 25));
+    child.kill('SIGINT');
+
     const exit = await waitForClose(child, 60_000);
     expect(
-      exit.signal === 'SIGINT' || exit.code === 130,
+      exit.signal === 'SIGINT' || exit.code === 130 || exit.code === 0,
       `unexpected exit: code=${exit.code}, signal=${exit.signal}`,
     ).toBe(true);
 

diff --git a/package-lock.json b/package-lock.json
diff --git a/package.json b/package.json
@@ -1,6 +1,6 @@
 {
   "name": "takt",
-  "version": "0.13.0",
+  "version": "0.14.0",
   "description": "TAKT: TAKT Agent Koordination Topology - AI Agent Piece Orchestration",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",
@@ -14,7 +14,7 @@
     "watch": "tsc --watch",
     "test": "vitest run",
     "test:watch": "vitest",
-    "test:e2e": "npm run test:e2e:mock; code=$?; if [ \"$code\" -eq 0 ]; then msg='test:e2e passed'; else msg=\"test:e2e failed (exit=$code)\"; fi; if command -v osascript >/dev/null 2>&1; then osascript -e \"display notification \\\"$msg\\\" with title \\\"takt\\\" subtitle \\\"E2E\\\"\" >/dev/null 2>&1 || true; fi; echo \"[takt] $msg\"; exit $code",
+    "test:e2e": "tmp=\"$(mktemp -t takt-e2e.XXXXXX)\"; npm run test:e2e:mock >\"$tmp\" 2>&1; code=$?; cat \"$tmp\"; if grep -q \"error connecting to api.github.com\" \"$tmp\"; then echo \"[takt] GitHub connectivity error detected in E2E output\"; code=1; fi; rm -f \"$tmp\"; if [ \"$code\" -eq 0 ]; then msg='test:e2e passed'; else msg=\"test:e2e failed (exit=$code)\"; fi; if command -v osascript >/dev/null 2>&1; then osascript -e \"display notification \\\"$msg\\\" with title \\\"takt\\\" subtitle \\\"E2E\\\"\" >/dev/null 2>&1 || true; fi; echo \"[takt] $msg\"; exit $code",
     "test:e2e:mock": "TAKT_E2E_PROVIDER=mock vitest run --config vitest.config.e2e.mock.ts --reporter=verbose",
     "test:e2e:all": "npm run test:e2e:mock && npm run test:e2e:provider",
     "test:e2e:provider": "npm run test:e2e:provider:claude && npm run test:e2e:provider:codex",

diff --git a/src/__tests__/actionDispatcher.test.ts b/src/__tests__/actionDispatcher.test.ts
@@ -0,0 +1,39 @@
+import { describe, it, expect, vi } from 'vitest';
+
+import { dispatchConversationAction } from '../features/interactive/actionDispatcher.js';
+
+describe('dispatchConversationAction', () => {
+  it('should dispatch to matching handler with full result payload', async () => {
+    const execute = vi.fn().mockResolvedValue('executed');
+    const saveTask = vi.fn().mockResolvedValue('saved');
+    const cancel = vi.fn().mockResolvedValue('cancelled');
+
+    const result = await dispatchConversationAction(
+      { action: 'save_task', task: 'refine branch docs' },
+      {
+        execute,
+        save_task: saveTask,
+        cancel,
+      },
+    );
+
+    expect(result).toBe('saved');
+    expect(saveTask).toHaveBeenCalledWith({ action: 'save_task', task: 'refine branch docs' });
+    expect(execute).not.toHaveBeenCalled();
+    expect(cancel).not.toHaveBeenCalled();
+  });
+
+  it('should support synchronous handlers', async () => {
+    const result = await dispatchConversationAction(
+      { action: 'cancel', task: '' },
+      {
+        execute: () => true,
+        save_task: () => true,
+        cancel: () => false,
+      },
+    );
+
+    expect(result).toBe(false);
+  });
+});
+