diff --git a/.github/workflows/readme-whats-new.yml b/.github/workflows/readme-whats-new.yml index 38d0d969..01d4e58f 100644 --- a/.github/workflows/readme-whats-new.yml +++ b/.github/workflows/readme-whats-new.yml @@ -251,13 +251,22 @@ jobs: mv /tmp/README_DE_updated.md docs/README_DE.md echo "docs/README_DE.md (DE) updated with new entry for $TODAY" - - name: Trim old entries + - name: Trim old entries and collapse if: steps.detect.outputs.has_features == 'true' run: | python3 <<'PYEOF' import re MAX_ENTRIES = 10 + VISIBLE_DATES = 2 # Number of most-recent unique dates shown expanded + + LANG_LABELS = { + "README.md": "Show earlier updates", + "docs/README_CN.md": "显示更早的更新", + "docs/README_JA.md": "以前の更新を表示", + "docs/README_FR.md": "Afficher les mises a jour precedentes", + "docs/README_DE.md": "Aeltere Updates anzeigen", + } for readme_file in [ "README.md", @@ -283,6 +292,14 @@ jobs: section = content[start_idx + len(start_marker):end_idx] after = content[end_idx:] + # Strip any existing
wrapper from previous runs + section = re.sub( + r'
\s*.*?\s*(.*?)\s*
', + r'\1', + section, + flags=re.DOTALL, + ) + # Split into date-headed entries entries = re.split(r'(?=\n#### \d{4}-\d{2}-\d{2})', section) entries = [e for e in entries if e.strip() and '####' in e] @@ -293,11 +310,33 @@ jobs: else: print(f"{readme_file}: Have {len(entries)} entries (max {MAX_ENTRIES}), no trimming needed.") - trimmed = "\n".join(entries) + "\n\n" - new_content = before + "\n" + trimmed + after + # Determine split point: first VISIBLE_DATES unique dates stay expanded + seen_dates = [] + split_idx = len(entries) + for i, entry in enumerate(entries): + m = re.search(r'#### (\d{4}-\d{2}-\d{2})', entry) + if m and m.group(1) not in seen_dates: + seen_dates.append(m.group(1)) + if len(seen_dates) > VISIBLE_DATES: + split_idx = i + break + + visible = entries[:split_idx] + collapsed = entries[split_idx:] + + result = "\n".join(visible) + if collapsed: + label = LANG_LABELS.get(readme_file, "Show earlier updates") + result += "\n\n
\n" + label + "\n" + result += "\n".join(collapsed) + result += "\n\n
\n" + result += "\n\n" + + new_content = before + "\n" + result + after with open(readme_file, "w", encoding="utf-8") as f: f.write(new_content) + print(f"{readme_file}: {len(visible)} entries visible, {len(collapsed)} collapsed.") PYEOF - name: Create branch, PR and merge diff --git a/README.md b/README.md index 6f7c168a..97b42008 100644 --- a/README.md +++ b/README.md @@ -61,12 +61,17 @@ It is built for researchers, developers, labs, and self-hosters who want more th - **Remote Profile Binding**: Deep research sessions can be bound to pre-configured SSH/remote compute profiles, enabling reproducible distributed research workflows + #### 2026-03-20 - **Deep Research Module**: Full AI-driven scientific research pipeline with multi-phase orchestration, reviewer deliberation, execution planning, and workflow graph UI - **Execution Pipeline**: Automated experiment execution system with Slurm job submission, dataset management, preprocessing, and remote executor support + +
+Show earlier updates + #### 2026-03-19 - **ClawHub Skill Import**: New integration to import skills directly from ClawHub via a dedicated API endpoint and import dialog - **Code Preview Panel**: New in-editor code preview component supporting syntax highlighting and save-status tracking @@ -75,6 +80,7 @@ It is built for researchers, developers, labs, and self-hosters who want more th + #### 2026-03-18 - **Multimodal Vision for Paper Analysis**: PDF images are now extracted and analyzed visually during paper discussion and research ideation sessions - **Claude Code Skills Integration**: Import skills directly from local folders or Claude Code projects via a new dedicated import workflow @@ -83,6 +89,7 @@ It is built for researchers, developers, labs, and self-hosters who want more th + #### 2026-03-18 - **Multimodal Vision for Paper Discussion & Ideation**: Vision-capable providers can now receive extracted PDF page images alongside text so discussion and ideation agents can analyze figures, tables, and diagrams. - **Paper Pages Gallery UI**: Discussion and ideation panels now show a collapsible thumbnail gallery for extracted paper pages with full-size preview dialogs. @@ -92,6 +99,7 @@ It is built for researchers, developers, labs, and self-hosters who want more th + #### 2026-03-17 - **Remote Job Profile Management & SSH Hardening**: Secure remote profile creation, editing, and SSH-hardened job submission for research execution - **Rich Markdown Rendering in Agent Panel**: Agent messages now render tables, LaTeX math, and syntax-highlighted code blocks @@ -102,6 +110,7 @@ It is built for researchers, developers, labs, and self-hosters who want more th + #### 2026-03-17 - **rjob Profile Config & Submission Hardening**: Remote profiles now store full rjob defaults (image, GPU, CPU, memory, mounts, charged-group, private-machine, env vars, host-network, example commands). `submitRemoteJob` builds the rjob command internally from stored config - the agent can no longer modify flags like `--charged-group` or `--image`. SSH transport fixed with `-o StrictHostKeyChecking=no -tt`, init script sourcing, and double-quote wrapping for correct quoting. - **Profile Editing**: Edit button (pencil icon) on remote profiles in the Remotes tab. Click to load profile into the form for updating, including all rjob config fields. @@ -112,6 +121,7 @@ It is built for researchers, developers, labs, and self-hosters who want more th + #### 2026-03-16 - **Paper Discussion & Ideation Robustness**: Per-role token budgets (2-2.5x increase), automatic retry on empty/short responses, and error visibility in the UI. Fixes agents returning empty or truncated output with reasoning-capable models (SH-Lab, Qwen, etc.) - **Full Paper Context**: Discussion and ideation agents now receive up to 30k chars of the full paper text (local files) instead of just the abstract, enabling deeper analysis of methodology, experiments, and results @@ -122,6 +132,7 @@ It is built for researchers, developers, labs, and self-hosters who want more th + #### 2026-03-14 - **Research Execution Engine**: New AI-driven research orchestration system with remote profiles, capability toggles, run history, and agent tools - **Auto-updating README "What's New"**: GitHub Actions workflow that automatically generates and commits a What's New section daily @@ -136,6 +147,11 @@ It is built for researchers, developers, labs, and self-hosters who want more th + + +
+ + --- diff --git a/docs/README_CN.md b/docs/README_CN.md index 0a7efa50..ea67afa9 100644 --- a/docs/README_CN.md +++ b/docs/README_CN.md @@ -51,12 +51,17 @@ InnoClaw 将服务器文件夹变成 AI 原生工作空间,用于基于文档 - **远程执行档案绑定**: 深度研究会话可绑定预配置的 SSH/远程计算档案,实现可复现的分布式研究工作流 + #### 2026-03-20 - **深度研究模块**: 完整的 AI 驱动科学研究流程,支持多阶段编排、评审员辩论、执行规划与工作流可视化界面 - **执行流水线**: 自动化实验执行系统,支持 Slurm 作业提交、数据集管理、预处理与远程执行器 + +
+显示更早的更新 + #### 2026-03-19 - **ClawHub 技能导入**: 新增从 ClawHub 直接导入技能的集成功能,包含专用 API 端点和导入对话框 - **代码预览面板**: 新增编辑器内代码预览组件,支持语法高亮及保存状态追踪 @@ -65,6 +70,7 @@ InnoClaw 将服务器文件夹变成 AI 原生工作空间,用于基于文档 + #### 2026-03-18 - **论文分析多模态视觉支持**: 在论文讨论和研究创意生成过程中,现可提取并视觉分析 PDF 中的图像内容 - **Claude Code 技能集成**: 新增专属导入流程,可直接从本地文件夹或 Claude Code 项目中导入技能 @@ -73,6 +79,7 @@ InnoClaw 将服务器文件夹变成 AI 原生工作空间,用于基于文档 + #### 2026-03-18 - **论文讨论与灵感生成支持多模态视觉**: 支持视觉的 AI 提供商现在会同时接收提取出的 PDF 页面图像和文本,使讨论与灵感生成智能体可以直接分析论文中的图表、表格与示意图。 - **论文页面图库界面**: Discussion 和 Ideation 面板新增可折叠的论文页面缩略图库,并支持点击查看大图预览。 @@ -82,6 +89,7 @@ InnoClaw 将服务器文件夹变成 AI 原生工作空间,用于基于文档 + #### 2026-03-17 - **远程作业配置管理与 SSH 安全加固**: 支持安全的远程配置文件创建、编辑及 SSH 加固的研究作业提交 - **智能体面板富文本渲染**: 智能体消息支持表格、LaTeX 数学公式及代码高亮渲染 @@ -92,6 +100,7 @@ InnoClaw 将服务器文件夹变成 AI 原生工作空间,用于基于文档 + #### 2026-03-17 - **rjob 配置与提交加固**: 远程配置现支持完整 rjob 默认值(镜像、GPU、CPU、内存、挂载、charged-group、私有机器、环境变量、host-network、示例命令)。`submitRemoteJob` 从存储配置内部构建 rjob 命令 - Agent 无法修改 `--charged-group` 或 `--image` 等参数。SSH 传输修复:`-o StrictHostKeyChecking=no -tt`、init 脚本加载及双引号包装。 - **远程配置编辑**: Remotes 标签页中远程配置新增编辑按钮(铅笔图标),点击可将配置加载到表单进行更新,包含所有 rjob 配置字段。 @@ -102,6 +111,7 @@ InnoClaw 将服务器文件夹变成 AI 原生工作空间,用于基于文档 + #### 2026-03-16 - **论文讨论与灵感生成稳定性提升**: 分角色 token 预算(提升 2-2.5 倍),空/短回复自动重试,UI 中显示错误信息。修复推理型模型(SH-Lab、Qwen 等)返回空或截断输出的问题 - **全文送入讨论智能体**: 讨论和灵感生成智能体现可接收最多 30k 字符的论文全文(本地文件),而非仅摘要,支持更深入的方法论、实验和结果分析 @@ -112,6 +122,7 @@ InnoClaw 将服务器文件夹变成 AI 原生工作空间,用于基于文档 + #### 2026-03-14 - **研究执行引擎**: 全新 AI 驱动的研究编排系统,支持远程配置、能力开关、运行历史和 Agent 工具 - **自动更新 README 新功能板块**: GitHub Actions 工作流每日自动生成并提交新功能板块 @@ -126,6 +137,11 @@ InnoClaw 将服务器文件夹变成 AI 原生工作空间,用于基于文档 + + +
+ + --- diff --git a/docs/README_DE.md b/docs/README_DE.md index d91f86df..cd380870 100644 --- a/docs/README_DE.md +++ b/docs/README_DE.md @@ -53,12 +53,17 @@ Es richtet sich an Forschende, Entwickler, Labore und Self-Hoster, die mehr als - **Entfernte Profil-Bindung**: Tiefe Recherche-Sitzungen koennen an vorkonfigurierte SSH/Remote-Rechenprofile gebunden werden, was reproduzierbare verteilte Forschungs-Workflows ermoeglicht + #### 2026-03-20 - **Tiefenforschungsmodul**: Vollstaendige KI-gesteuerte wissenschaftliche Forschungs-Pipeline mit Mehrphasen-Orchestrierung, Gutachter-Diskussion, Ausfuehrungsplanung und Workflow-Grafik-Oberflaeche - **Ausfuehrungs-Pipeline**: Automatisiertes System zur Experimentausfuehrung mit Slurm-Job-Uebermittlung, Datensatzverwaltung, Vorverarbeitung und Unterstuetzung fuer entfernte Ausfuehrer + +
+Aeltere Updates anzeigen + #### 2026-03-19 - **ClawHub-Skill-Import**: Neue Integration zum direkten Importieren von Skills aus ClawHub ueber einen dedizierten API-Endpunkt und einen Import-Dialog - **Code-Vorschaufenster**: Neue In-Editor-Komponente fuer die Code-Vorschau mit Syntaxhervorhebung und Verfolgung des Speicherstatus @@ -67,6 +72,7 @@ Es richtet sich an Forschende, Entwickler, Labore und Self-Hoster, die mehr als + #### 2026-03-18 - **Multimodale Bildanalyse fuer Papierauswertung**: PDF-Bilder werden jetzt waehrend Diskussions- und Forschungsideensitzungen visuell extrahiert und analysiert - **Claude Code Skills-Integration**: Importieren Sie Skills direkt aus lokalen Ordnern oder Claude Code-Projekten ueber einen neuen dedizierten Import-Workflow @@ -75,6 +81,7 @@ Es richtet sich an Forschende, Entwickler, Labore und Self-Hoster, die mehr als + #### 2026-03-18 - **Multimodale Vision fuer Paper-Diskussion und Ideation**: Vision-faehige Anbieter erhalten jetzt extrahierte PDF-Seitenbilder zusammen mit Text, damit Agents Abbildungen, Tabellen und Diagramme direkt analysieren koennen. - **Paper-Seitengalerie-UI**: Die Discussion- und Ideation-Panels zeigen jetzt eine einklappbare Miniaturgalerie der extrahierten Paper-Seiten mit Grossansicht im Dialog. @@ -84,6 +91,7 @@ Es richtet sich an Forschende, Entwickler, Labore und Self-Hoster, die mehr als + #### 2026-03-17 - **Remote-Job-Profilverwaltung und SSH-Haertung**: sichere Erstellung, Bearbeitung und SSH-gehaertete Einreichung von Forschungsjobs auf Remote-Systemen - **Rich Markdown Rendering im Agent Panel**: Agent-Nachrichten rendern jetzt Tabellen, LaTeX-Mathematik und syntaxhervorgehobene Codebloecke @@ -93,6 +101,7 @@ Es richtet sich an Forschende, Entwickler, Labore und Self-Hoster, die mehr als + #### 2026-03-17 - **rjob-Profilkonfiguration und sichere Einreichung**: Remote-Profile speichern vollstaendige rjob-Defaults wie image, GPU, CPU, memory, mounts, charged-group, private-machine, env vars, host-network und example commands. `submitRemoteJob` baut den rjob-Befehl intern aus der gespeicherten Konfiguration auf, sodass der Agent kritische Flags wie `--charged-group` oder `--image` nicht veraendern kann. Auch der SSH-Transport wurde mit `-o StrictHostKeyChecking=no -tt`, dem Laden des Init-Skripts und korrektem Quoting verhaertet. - **Profilbearbeitung**: Der Edit-Button im Remotes-Tab laedt vorhandene Profile inklusive aller rjob-Felder zur Aktualisierung in das Formular. @@ -102,6 +111,7 @@ Es richtet sich an Forschende, Entwickler, Labore und Self-Hoster, die mehr als + #### 2026-03-16 - **Robustheit fuer Paper-Diskussion und Ideation**: 2-2.5x hoeheres Token-Budget pro Rolle, automatische Wiederholung bei leeren oder zu kurzen Antworten und sichtbare Fehler in der UI - **Vollstaendiger Paper-Kontext**: Discussion- und Ideation-Agents erhalten bis zu 30k Zeichen des lokalen Volltexts statt nur des Abstracts @@ -111,6 +121,7 @@ Es richtet sich an Forschende, Entwickler, Labore und Self-Hoster, die mehr als + #### 2026-03-14 - **Research Execution Engine**: neues KI-gesteuertes Forschungsorchestrierungssystem mit Remote-Profilen, Capability-Toggles, Laufhistorie und Agent-Tools - **Automatisch aktualisierter README-Bereich "What's New"**: GitHub-Actions-Workflow, der wichtige neue Features taeglich erkennt und in die README eintraegt @@ -123,6 +134,11 @@ Es richtet sich an Forschende, Entwickler, Labore und Self-Hoster, die mehr als + + +
+ + --- diff --git a/docs/README_FR.md b/docs/README_FR.md index b12ce95d..784b73c1 100644 --- a/docs/README_FR.md +++ b/docs/README_FR.md @@ -53,12 +53,17 @@ Il s'adresse aux chercheurs, developpeurs, laboratoires et adeptes du self-hosti - **Liaison de profil distant**: Les sessions de recherche approfondie peuvent etre liees a des profils SSH/calcul distant preconfigures, permettant des workflows de recherche distribues et reproductibles + #### 2026-03-20 - **Module de recherche approfondie**: Pipeline de recherche scientifique entierement pilote par IA avec orchestration multi-phases, deliberation des evaluateurs, planification d'execution et interface graphique de flux de travail - **Pipeline d'execution**: Systeme d'execution d'experiences automatise avec soumission de jobs Slurm, gestion de jeux de donnees, preprocessement et support d'executeurs distants + +
+Afficher les mises a jour precedentes + #### 2026-03-19 - **Importation de competences ClawHub**: Nouvelle integration pour importer des competences directement depuis ClawHub via un point d'API dedie et une boite de dialogue d'importation - **Panneau de previsualisation du code**: Nouveau composant de previsualisation de code integre a l'editeur, avec coloration syntaxique et suivi de l'etat de sauvegarde @@ -67,6 +72,7 @@ Il s'adresse aux chercheurs, developpeurs, laboratoires et adeptes du self-hosti + #### 2026-03-18 - **Vision Multimodale pour l'Analyse d'Articles**: Les images PDF sont desormais extraites et analysees visuellement lors des sessions de discussion et d'ideation de recherche - **Integration des Competences Claude Code**: Importez des competences directement depuis des dossiers locaux ou des projets Claude Code via un nouveau flux d'importation dedie @@ -75,6 +81,7 @@ Il s'adresse aux chercheurs, developpeurs, laboratoires et adeptes du self-hosti + #### 2026-03-18 - **Vision multimodale pour discussion et ideation d'articles** : les fournisseurs compatibles vision recoivent maintenant les images de pages PDF extraites en plus du texte afin d'analyser figures, tableaux et schemas. - **UI de galerie des pages d'article** : les panneaux Discussion et Ideation affichent maintenant une galerie repliable de miniatures de pages avec apercu en grand format. @@ -84,6 +91,7 @@ Il s'adresse aux chercheurs, developpeurs, laboratoires et adeptes du self-hosti + #### 2026-03-17 - **Gestion des profils de jobs distants et durcissement SSH** : prise en charge de la creation, de l'edition et de l'envoi securise de jobs de recherche via SSH - **Rendu Markdown riche dans le panneau Agent** : les messages Agent affichent maintenant tableaux, formules LaTeX et blocs de code avec coloration syntaxique @@ -93,6 +101,7 @@ Il s'adresse aux chercheurs, developpeurs, laboratoires et adeptes du self-hosti + #### 2026-03-17 - **Durcissement de la configuration et de la soumission rjob** : les profils distants peuvent stocker les valeurs par defaut rjob completes (image, GPU, CPU, memory, mounts, charged-group, private-machine, env vars, host-network, example commands). `submitRemoteJob` construit la commande rjob a partir de la configuration stockee, ce qui empeche l'Agent de modifier des drapeaux critiques comme `--charged-group` ou `--image`. Le transport SSH a aussi ete fiabilise avec `-o StrictHostKeyChecking=no -tt`, le chargement du script d'init et le bon quoting. - **Edition des profils** : le bouton d'edition dans l'onglet Remotes recharge le profil existant dans le formulaire, y compris tous les champs rjob. @@ -102,6 +111,7 @@ Il s'adresse aux chercheurs, developpeurs, laboratoires et adeptes du self-hosti + #### 2026-03-16 - **Robustesse des discussions d'articles et de l'ideation** : budget de tokens par role augmente de 2 a 2.5x, reessai automatique sur reponses vides ou trop courtes, et erreurs visibles dans l'UI - **Contexte article complet** : les agents de discussion et d'ideation recoivent jusqu'a 30k caracteres du texte complet local de l'article, et pas seulement le resume @@ -111,6 +121,7 @@ Il s'adresse aux chercheurs, developpeurs, laboratoires et adeptes du self-hosti + #### 2026-03-14 - **Research Execution Engine** : nouveau systeme d'orchestration de recherche pilote par IA avec profils distants, toggles de capacites, historique d'execution et outils Agent - **Section README "What's New" mise a jour automatiquement** : workflow GitHub Actions qui genere et met a jour chaque jour les nouvelles fonctionnalites importantes @@ -123,6 +134,11 @@ Il s'adresse aux chercheurs, developpeurs, laboratoires et adeptes du self-hosti + + +
+ + --- diff --git a/docs/README_JA.md b/docs/README_JA.md index d0117a31..cd282ff2 100644 --- a/docs/README_JA.md +++ b/docs/README_JA.md @@ -53,12 +53,17 @@ InnoClaw はサーバー上のフォルダを、文書に根ざした対話、 - **リモートプロファイルバインディング**: 深層調査セッションを事前設定済みの SSH/リモート計算プロファイルに紐付け、再現可能な分散型研究ワークフローを実現 + #### 2026-03-20 - **深層リサーチモジュール**: 多段階オーケストレーション、レビュアー討議、実行計画、ワークフローグラフUIを備えた完全AI驱動の科学研究パイプライン - **実行パイプライン**: Slurmジョブ送信、データセット管理、前処理、リモート実行器をサポートする自動化実験実行システム + +
+以前の更新を表示 + #### 2026-03-19 - **ClawHub スキルインポート**: ClawHub から専用 API エンドポイントとインポートダイアログを通じて、スキルを直接インポートできる新しい連携機能 - **コードプレビューパネル**: シンタックスハイライトと保存状態のトラッキングに対応した、エディタ内コードプレビューコンポーネントを新規追加 @@ -67,6 +72,7 @@ InnoClaw はサーバー上のフォルダを、文書に根ざした対話、 + #### 2026-03-18 - **論文分析のマルチモーダルビジョン**: 論文ディスカッションおよびリサーチアイデア出しセッション中に、PDF 画像を抽出してビジュアル分析できるようになりました - **Claude Code スキル統合**: 新しい専用インポートワークフローにより、ローカルフォルダーまたは Claude Code プロジェクトからスキルを直接インポートできます @@ -75,6 +81,7 @@ InnoClaw はサーバー上のフォルダを、文書に根ざした対話、 + #### 2026-03-18 - **論文討論とアイデア生成のマルチモーダル視覚対応**: 視覚対応プロバイダーでは、抽出した PDF ページ画像が本文テキストと一緒に渡され、図、表、ダイアグラムを直接解析できます。 - **論文ページギャラリー UI**: Discussion と Ideation パネルに折りたたみ式のページ縮略图库が追加され、クリックで大きなプレビューを表示できます。 @@ -84,6 +91,7 @@ InnoClaw はサーバー上のフォルダを、文書に根ざした対話、 + #### 2026-03-17 - **リモートジョブプロファイル管理と SSH 強化**: 研究実行向けに、安全なリモートプロファイル作成・編集と SSH 強化済みのジョブ送信をサポート - **Agent パネルのリッチ Markdown 描画**: Agent メッセージで表、LaTeX 数式、シンタックスハイライト付きコードブロックを表示 @@ -93,6 +101,7 @@ InnoClaw はサーバー上のフォルダを、文書に根ざした対話、 + #### 2026-03-17 - **rjob プロファイル設定と送信の強化**: リモートプロファイルに image、GPU、CPU、memory、mounts、charged-group、private-machine、env vars、host-network、example commands などの rjob 既定値を保存可能。`submitRemoteJob` は保存済み設定から内部的に rjob コマンドを構築するため、Agent は `--charged-group` や `--image` などの重要フラグを変更できません。SSH 転送も `-o StrictHostKeyChecking=no -tt`、init スクリプト読込、二重引用で改善しました。 - **プロファイル編集**: Remotes タブの既存リモートプロファイルに編集ボタンを追加。rjob 関連項目を含む全フィールドをフォームに読み込んで更新できます。 @@ -102,6 +111,7 @@ InnoClaw はサーバー上のフォルダを、文書に根ざした対話、 + #### 2026-03-16 - **論文討論とアイデア生成の堅牢化**: 役割ごとの token 予算を 2-2.5 倍に増やし、空または短い応答を自動再試行、UI にエラーも表示。推論系モデルで空応答や切り詰めが起きる問題を改善 - **論文全文コンテキスト**: Discussion と Ideation の Agent が要約だけでなく、ローカル論文本文を最大 30k 文字まで受け取り、方法・実験・結果をより深く分析可能 @@ -111,6 +121,7 @@ InnoClaw はサーバー上のフォルダを、文書に根ざした対話、 + #### 2026-03-14 - **Research Execution Engine**: リモートプロファイル、能力トグル、実行履歴、Agent ツールを備えた新しい AI 主導の研究オーケストレーション機能 - **README の自動更新される "What's New"**: GitHub Actions ワークフローが毎日の重要な新機能を自動生成して README に反映 @@ -123,6 +134,11 @@ InnoClaw はサーバー上のフォルダを、文書に根ざした対話、 + + +
+ + ---