From cf97c629590e4a3e1d4462a9ebea7f767d4c40a2 Mon Sep 17 00:00:00 2001 From: nrslib <38722970+nrslib@users.noreply.github.com> Date: Sun, 8 Feb 2026 20:33:24 +0900 Subject: [PATCH] Release v0.9.0 --- CHANGELOG.md | 35 +++++++++++++++++++++++++++++++++++ README.md | 18 ++++++++++++++++-- docs/README.ja.md | 18 ++++++++++++++++-- package-lock.json | 4 ++-- package.json | 2 +- 5 files changed, 70 insertions(+), 7 deletions(-) diff --git a/CHANGELOG.md b/CHANGELOG.md index 62ab39a..4321142 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -4,6 +4,41 @@ 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/). +## [0.9.0] - 2026-02-08 + +### Added + +- **`takt catalog` command**: List available facets (personas, policies, knowledge, instructions, output-contracts) across layers (builtin/user/project) +- **`compound-eye` builtin piece**: Multi-model review — sends the same instruction to Claude and Codex simultaneously, then synthesizes both responses +- **Parallel task execution**: `takt run` now uses a worker pool for concurrent task execution (controlled by `concurrency` config, default: 1) +- **Rich line editor in interactive mode**: Shift+Enter for multiline input, cursor movement (arrow keys, Home/End), Option+Arrow word movement, Ctrl+A/E/K/U/W editing, paste bracket mode support +- **Movement preview in interactive mode**: Injects piece movement structure (persona + instruction) into the AI planner for improved task analysis (`interactive_preview_movements` config, default: 3) +- **MCP server configuration**: Per-movement MCP (Model Context Protocol) server settings with stdio/SSE/HTTP transport support +- **Facet-level eject**: `takt eject persona coder` — eject individual facets by type and name for customization +- **3-layer facet resolution**: Personas, policies, and other facets resolved via project → user → builtin lookup (name-based references supported) +- **`pr-commenter` persona**: Specialized persona for posting review findings as GitHub PR comments +- **`notification_sound` config**: Enable/disable notification sounds (default: true) +- **Prompt log viewer**: `tools/prompt-log-viewer.html` for visualizing prompt-response pairs during debugging +- Auto-PR base branch now set to the current branch before branch creation + +### Changed + +- Unified planner and architect-planner: extracted design knowledge into knowledge facets, merged into planner. Removed architect movement from default/coding pieces (plan → implement direct transition) +- Replaced readline with raw-mode line editor in interactive mode (cursor management, inter-line movement, Kitty keyboard protocol) +- Unified interactive mode `save_task` with `takt add` worktree setup flow +- Added `-d` flag to caffeinate to prevent App Nap process freezing during display sleep +- Issue references now routed through interactive mode (previously executed directly, now used as initial input) +- SDK update: `@anthropic-ai/claude-agent-sdk` v0.2.34 → v0.2.37 +- Enhanced interactive session scoring prompts with piece structure information + +### Internal + +- Extracted `resource-resolver.ts` for facet resolution logic (separated from `pieceParser.ts`) +- Extracted `parallelExecution.ts` (worker pool), `resolveTask.ts` (task resolution), `sigintHandler.ts` (shared SIGINT handler) +- Unified session key generation via `session-key.ts` +- New `lineEditor.ts` (raw-mode terminal input, escape sequence parsing, cursor management) +- Extensive test additions: catalog, facet-resolution, eject-facet, lineEditor, formatMovementPreviews, models, debug, strip-ansi, workerPool, runAllTasks-concurrency, session-key, interactive (major expansion), cli-routing-issue-resolve, parallel-logger, engine-parallel-failure, StreamDisplay, getCurrentBranch, globalConfig-defaults, pieceExecution-debug-prompts, selectAndExecute-autoPr, it-notification-sound, it-piece-loader, permission-mode (expansion) + ## [0.8.0] - 2026-02-08 alpha.1 の内容を正式リリース。機能変更なし。 diff --git a/README.md b/README.md index d5e1e28..59c39ba 100644 --- a/README.md +++ b/README.md @@ -262,6 +262,14 @@ takt clear # Deploy builtin pieces/personas as Claude Code Skill takt export-cc +# List available facets across layers +takt catalog +takt catalog personas + +# Eject a specific facet for customization +takt eject persona coder +takt eject instruction plan --global + # Preview assembled prompts for each movement and phase takt prompt [piece] @@ -432,15 +440,16 @@ TAKT includes multiple builtin pieces: | Piece | Description | |----------|-------------| -| `default` | Full development piece: plan → architecture design → implement → AI review → parallel review (architect + security) → supervisor approval. Includes fix loops at each review stage. | +| `default` | Full development piece: plan → implement → AI review → parallel review (architect + QA) → supervisor approval. Includes fix loops at each review stage. | | `minimal` | Quick piece: plan → implement → review → supervisor. Minimal steps for fast iteration. | | `review-fix-minimal` | Review-focused piece: review → fix → supervisor. For iterative improvement based on review feedback. | | `research` | Research piece: planner → digger → supervisor. Autonomously executes research without asking questions. | | `expert` | Full-stack development piece: architecture, frontend, security, QA reviews with fix loops. | | `expert-cqrs` | Full-stack development piece (CQRS+ES specialized): CQRS+ES, frontend, security, QA reviews with fix loops. | | `magi` | Deliberation system inspired by Evangelion. Three AI personas (MELCHIOR, BALTHASAR, CASPER) analyze and vote. | -| `coding` | Lightweight development piece: architect-planner → implement → parallel review (AI antipattern + architecture) → fix. Fast feedback loop without supervisor. | +| `coding` | Lightweight development piece: planner → implement → parallel review (AI antipattern + architecture) → fix. Fast feedback loop without supervisor. | | `passthrough` | Thinnest wrapper. Pass task directly to coder as-is. No review. | +| `compound-eye` | Multi-model review: sends the same instruction to Claude and Codex simultaneously, then synthesizes both responses. | | `review-only` | Read-only code review piece that makes no changes. | **Hybrid Codex variants** (`*-hybrid-codex`): Each major piece has a Codex variant where the coder agent runs on Codex while reviewers use Claude. Available for: default, minimal, expert, expert-cqrs, passthrough, review-fix-minimal, coding. @@ -466,6 +475,7 @@ Use `takt switch` to switch pieces. | **research-planner** | Research task planning and scope definition | | **research-digger** | Deep investigation and information gathering | | **research-supervisor** | Research quality validation and completeness assessment | +| **pr-commenter** | Posts review findings as GitHub PR comments | ## Custom Personas @@ -531,6 +541,9 @@ provider: claude # Default provider: claude or codex model: sonnet # Default model (optional) branch_name_strategy: romaji # Branch name generation: 'romaji' (fast) or 'ai' (slow) prevent_sleep: false # Prevent macOS idle sleep during execution (caffeinate) +notification_sound: true # Enable/disable notification sounds +concurrency: 1 # Parallel task count for takt run (1-10, default: 1 = sequential) +interactive_preview_movements: 3 # Movement previews in interactive mode (0-10, default: 3) # API Key configuration (optional) # Can be overridden by environment variables TAKT_ANTHROPIC_API_KEY / TAKT_OPENAI_API_KEY @@ -746,6 +759,7 @@ Special `next` values: `COMPLETE` (success), `ABORT` (failure) | `permission_mode` | - | Permission mode: `readonly`, `edit`, `full` (provider-independent) | | `output_contracts` | - | Output contract definitions for report files | | `quality_gates` | - | AI directives for movement completion requirements | +| `mcp_servers` | - | MCP (Model Context Protocol) server configuration (stdio/SSE/HTTP) | ## API Usage Example diff --git a/docs/README.ja.md b/docs/README.ja.md index 535e8ee..8121444 100644 --- a/docs/README.ja.md +++ b/docs/README.ja.md @@ -258,6 +258,14 @@ takt clear # ビルトインピース・エージェントを Claude Code Skill としてデプロイ takt export-cc +# 利用可能なファセットをレイヤー別に一覧表示 +takt catalog +takt catalog personas + +# 特定のファセットをカスタマイズ用にコピー +takt eject persona coder +takt eject instruction plan --global + # 各ムーブメント・フェーズの組み立て済みプロンプトをプレビュー takt prompt [piece] @@ -428,15 +436,16 @@ TAKTには複数のビルトインピースが同梱されています: | ピース | 説明 | |------------|------| -| `default` | フル開発ピース: 計画 → アーキテクチャ設計 → 実装 → AI レビュー → 並列レビュー（アーキテクト＋セキュリティ）→ スーパーバイザー承認。各レビュー段階に修正ループあり。 | +| `default` | フル開発ピース: 計画 → 実装 → AI レビュー → 並列レビュー（アーキテクト＋QA）→ スーパーバイザー承認。各レビュー段階に修正ループあり。 | | `minimal` | クイックピース: 計画 → 実装 → レビュー → スーパーバイザー。高速イテレーション向けの最小構成。 | | `review-fix-minimal` | レビュー重視ピース: レビュー → 修正 → スーパーバイザー。レビューフィードバックに基づく反復改善向け。 | | `research` | リサーチピース: プランナー → ディガー → スーパーバイザー。質問せずに自律的にリサーチを実行。 | | `expert` | フルスタック開発ピース: アーキテクチャ、フロントエンド、セキュリティ、QA レビューと修正ループ。 | | `expert-cqrs` | フルスタック開発ピース（CQRS+ES特化）: CQRS+ES、フロントエンド、セキュリティ、QA レビューと修正ループ。 | | `magi` | エヴァンゲリオンにインスパイアされた審議システム。3つの AI ペルソナ（MELCHIOR、BALTHASAR、CASPER）が分析し投票。 | -| `coding` | 軽量開発ピース: architect-planner → 実装 → 並列レビュー（AI アンチパターン＋アーキテクチャ）→ 修正。スーパーバイザーなしの高速フィードバックループ。 | +| `coding` | 軽量開発ピース: planner → 実装 → 並列レビュー（AI アンチパターン＋アーキテクチャ）→ 修正。スーパーバイザーなしの高速フィードバックループ。 | | `passthrough` | 最小構成。タスクをそのまま coder に渡す薄いラッパー。レビューなし。 | +| `compound-eye` | マルチモデルレビュー: Claude と Codex に同じ指示を同時送信し、両方の回答を統合。 | | `review-only` | 変更を加えない読み取り専用のコードレビューピース。 | **Hybrid Codex バリアント** (`*-hybrid-codex`): 主要ピースごとに、coder エージェントを Codex で実行しレビュアーは Claude を使うハイブリッド構成が用意されています。対象: default, minimal, expert, expert-cqrs, passthrough, review-fix-minimal, coding。 @@ -462,6 +471,7 @@ TAKTには複数のビルトインピースが同梱されています: | **research-planner** | リサーチタスクの計画・スコープ定義 | | **research-digger** | 深掘り調査と情報収集 | | **research-supervisor** | リサーチ品質の検証と網羅性の評価 | +| **pr-commenter** | レビュー結果を GitHub PR にコメントとして投稿 | ## カスタムペルソナ @@ -527,6 +537,9 @@ provider: claude # デフォルトプロバイダー: claude または c model: sonnet # デフォルトモデル（オプション） branch_name_strategy: romaji # ブランチ名生成: 'romaji'（高速）または 'ai'（低速） prevent_sleep: false # macOS の実行中スリープ防止（caffeinate） +notification_sound: true # 通知音の有効/無効 +concurrency: 1 # takt run の並列タスク数（1-10、デフォルト: 1 = 逐次実行） +interactive_preview_movements: 3 # 対話モードでのムーブメントプレビュー数（0-10、デフォルト: 3） # API Key 設定（オプション） # 環境変数 TAKT_ANTHROPIC_API_KEY / TAKT_OPENAI_API_KEY で上書き可能 @@ -742,6 +755,7 @@ rules: | `permission_mode` | - | パーミッションモード: `readonly`、`edit`、`full`（プロバイダー非依存） | | `output_contracts` | - | レポートファイルの出力契約定義 | | `quality_gates` | - | ムーブメント完了要件のAIディレクティブ | +| `mcp_servers` | - | MCP（Model Context Protocol）サーバー設定（stdio/SSE/HTTP） | ## API使用例 diff --git a/package-lock.json b/package-lock.json index da99bef..f0ab485 100644 --- a/package-lock.json +++ b/package-lock.json @@ -1,12 +1,12 @@ { "name": "takt", - "version": "0.8.0", + "version": "0.9.0", "lockfileVersion": 3, "requires": true, "packages": { "": { "name": "takt", - "version": "0.8.0", + "version": "0.9.0", "license": "MIT", "dependencies": { "@anthropic-ai/claude-agent-sdk": "^0.2.37", diff --git a/package.json b/package.json index fb6b109..46bd391 100644 --- a/package.json +++ b/package.json @@ -1,6 +1,6 @@ { "name": "takt", - "version": "0.8.0", + "version": "0.9.0", "description": "TAKT: Task Agent Koordination Tool - AI Agent Piece Orchestration", "main": "dist/index.js", "types": "dist/index.d.ts",