From aa372a8fd55eaaf932a4ea625404ed194af27706 Mon Sep 17 00:00:00 2001 From: catlog22 Date: Mon, 3 Nov 2025 17:42:00 +0800 Subject: [PATCH] docs: correct CHANGELOG.md for v5.2.0 - highlight /memory:skill-memory command MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit - Update v5.2.0 focus from /memory:docs to /memory:skill-memory (actual new feature) - Add comprehensive 4-phase orchestrator workflow description - Document progressive loading levels (Level 0-3, 2K-40K tokens) - Include path mirroring strategy and SKILL package structure - Highlight /memory:docs enhancements as secondary improvements - Add usage examples and output format 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude --- CHANGELOG.md | 176 +++++++++++++++++++++++++++++++-------------------- 1 file changed, 106 insertions(+), 70 deletions(-) diff --git a/CHANGELOG.md b/CHANGELOG.md index 1d716c93..3a39ebbb 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -7,105 +7,141 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0 ## [5.2.0] - 2025-11-03 -### 🎉 New Command: `/memory:docs` - Intelligent Documentation Workflow +### 🎉 New Command: `/memory:skill-memory` - SKILL Package Generator -This release introduces a powerful new documentation command that revolutionizes project documentation generation with intelligent planning, parallel execution, and smart task grouping. +This release introduces a powerful new command that automatically generates progressive-loading SKILL packages from project documentation with intelligent orchestration and path mirroring. #### ✅ Added -**New `/memory:docs` Command**: -- ✨ **Lightweight Documentation Planner** - Analyzes project structure and generates executable documentation tasks -- ✨ **Smart Task Grouping** - Dynamic grouping algorithm with ≤7 documents per task constraint -- ✨ **Context Sharing Optimization** - Prefer grouping 2 top-level directories for shared Gemini analysis -- ✨ **Parallel Execution** - Multiple task groups execute concurrently for faster completion -- ✨ **Path Mirroring Strategy** - Documentation structure mirrors source code hierarchy -- ✨ **Two Execution Modes** - Agent Mode (default) and CLI Mode (--cli-execute) -- ✨ **Two Documentation Modes** - Full mode (complete docs) and Partial mode (modules only) - -**Task Grouping Algorithm**: -- 🎯 **Primary constraint**: Each task generates ≤7 documents (hard limit) -- 🎯 **Optimization goal**: Group 2 directories when possible for context sharing -- 🎯 **Conflict resolution**: Automatic splitting when exceeding document limit -- 🎯 **Context benefit**: Same-task directories analyzed via single Gemini call - -**Workflow Phases**: -- 📋 **Phase 1**: Initialize session and create directory structure -- 📋 **Phase 2**: Pre-computed analysis (folder classification, structure analysis) -- 📋 **Phase 3**: Detect update mode (create vs. update) -- 📋 **Phase 4**: Smart task decomposition with document count control -- 📋 **Phase 5**: Generate executable task JSONs with dependency chains - -**Documentation Types**: -- 📚 **Module Documentation**: API.md (code folders) + README.md (all folders) -- 📚 **Project Documentation**: README.md (overview and navigation) -- 📚 **Architecture Documentation**: ARCHITECTURE.md + EXAMPLES.md -- 📚 **API Documentation**: HTTP API reference (optional, auto-detected) - -**Intelligent Features**: -- 🧠 **Auto-detection** - Skip tests/build/config/vendor directories automatically -- 🧠 **Folder Classification** - Code folders vs. Navigation folders -- 🧠 **Incremental Updates** - Preserve user modifications in existing docs -- 🧠 **Pre-computed Analysis** - Phase 2 analysis stored in `.process/`, reused across tasks -- 🧠 **Efficient Data Loading** - Existing docs loaded once, shared across all tasks +**New `/memory:skill-memory` Command**: +- ✨ **4-Phase Orchestrator** - Automated workflow from documentation to SKILL package + - Phase 1: Parse arguments and prepare environment + - Phase 2: Call `/memory:docs` to plan documentation + - Phase 3: Call `/workflow:execute` to generate documentation + - Phase 4: Generate SKILL.md index with progressive loading +- ✨ **Auto-Continue Mechanism** - All phases run autonomously via TodoList tracking +- ✨ **Path Mirroring** - SKILL knowledge structure mirrors source code hierarchy +- ✨ **Progressive Loading** - 4-level token-budgeted documentation access + - Level 0: Quick Start (~2K tokens) - README only + - Level 1: Core Modules (~8K tokens) - Module READMEs + - Level 2: Complete (~25K tokens) - All modules + Architecture + - Level 3: Deep Dive (~40K tokens) - Everything + Examples +- ✨ **Intelligent Description Generation** - Auto-extracts capabilities and triggers from documentation +- ✨ **Regeneration Support** - `--regenerate` flag to force fresh documentation +- ✨ **Multi-Tool Support** - Supports gemini, qwen, and codex for documentation generation **Command Parameters**: ```bash -/memory:docs [path] [--tool ] [--mode ] [--cli-execute] +/memory:skill-memory [path] [--tool ] [--regenerate] [--mode ] [--cli-execute] ``` -#### 🔧 Technical Highlights +**Path Mirroring Strategy**: +``` +Source: my_app/src/modules/auth/ + ↓ +Docs: .workflow/docs/my_app/src/modules/auth/API.md + ↓ +SKILL: .claude/skills/my_app/knowledge/src/modules/auth/API.md +``` -**Session Structure**: -- `.workflow/WFS-docs-{timestamp}/` - Session root directory -- `.process/phase2-analysis.json` - Single JSON with all pre-computed analysis -- `.task/IMPL-*.json` - Executable task definitions with dependency chains +**4-Phase Workflow**: +1. **Prepare**: Parse arguments, check existing docs, handle --regenerate +2. **Plan**: Call `/memory:docs` to create documentation tasks +3. **Execute**: Call `/workflow:execute` to generate documentation files +4. **Index**: Generate SKILL.md with progressive loading structure -**Task Hierarchy** (Dynamic based on document count): -- **Level 1**: Module tree groups (parallel execution, ≤7 docs each) -- **Level 2**: Project README (depends on Level 1, full mode only) -- **Level 3**: ARCHITECTURE + EXAMPLES (depends on Level 2, full mode only) -- **Level 4**: HTTP API documentation (optional, auto-detected) +**SKILL Package Output**: +- `.claude/skills/{project_name}/SKILL.md` - Index with progressive loading levels +- `.claude/skills/{project_name}/knowledge/` - Mirrored documentation structure +- Automatic capability detection and trigger phrase generation -**Grouping Examples**: -- Small projects (≤7 docs): Single task with shared context -- Medium projects (>7 docs): Multiple groups with 2 dirs each -- Large projects (single dir >7 docs): Automatic subdirectory splitting +#### 📝 Changed -#### 🎯 Performance Benefits +**Enhanced `/memory:docs` Command**: +- 🔄 **Smart Task Grouping** - ≤7 documents per task (up from 5) +- 🔄 **Context Sharing** - Prefer grouping 2 top-level directories for shared Gemini analysis +- 🔄 **Batch Processing** - Reduced task count through intelligent grouping +- 🔄 **Dual Execution Modes** - Agent Mode (default) and CLI Mode (--cli-execute) +- 🔄 **Pre-computed Analysis** - Phase 2 unified analysis eliminates redundant CLI calls +- 🔄 **Conflict Resolution** - Automatic splitting when exceeding document limit -**Resource Optimization**: +**Documentation Workflow Improvements**: +- 🔄 **CLI Execute Support** - Direct documentation generation via CLI tools (gemini/qwen/codex) +- 🔄 **workflow-session.json** - Unified session metadata storage +- 🔄 **Improved Structure Quality** - Enhanced documentation generation guidelines + +#### 🎯 Benefits + +**SKILL Package Features**: +- 📦 **Progressive Loading** - Load only what you need (2K → 40K tokens) +- 📦 **Path Mirroring** - Easy navigation matching source structure +- 📦 **Auto-Discovery** - Intelligent capability and trigger detection +- 📦 **Regeneration** - Force fresh docs with single flag +- 📦 **Zero Manual Steps** - Fully automated 4-phase workflow + +**Performance Optimization**: - ⚡ **Parallel Processing** - Multiple directory groups execute concurrently - ⚡ **Context Sharing** - Single Gemini call per task group (2 directories) - ⚡ **Efficient Analysis** - One-time analysis in Phase 2, reused by all tasks - ⚡ **Predictable Sizing** - ≤7 docs per task ensures reliable completion - ⚡ **Failure Isolation** - Task-level failures don't block entire workflow -**Data Efficiency**: -- 📊 Single `phase2-analysis.json` replaces 7+ temporary files -- 📊 Existing docs loaded once, shared across all tasks -- 📊 Pre-computed folder analysis eliminates redundant scanning +**Workflow Integration**: +- 🔗 Seamless integration with existing `/memory:docs` command +- 🔗 Compatible with `/workflow:execute` system +- 🔗 Auto-continue mechanism eliminates manual steps +- 🔗 TodoList progress tracking throughout workflow -#### 📦 New Files +#### 📦 New/Modified Files -- `.claude/commands/memory/docs.md` - Complete command specification and workflow -- Integration with existing `@doc-generator` agent -- Compatible with workflow execution system (`/workflow:execute`) +**New**: +- `.claude/commands/memory/skill-memory.md` - Complete command specification (822 lines) -#### 🔗 Integration +**Modified**: +- `.claude/commands/memory/docs.md` - Enhanced with batch processing and smart grouping +- `.claude/agents/doc-generator.md` - Mode-aware execution support -**Execute documentation workflow**: +#### 🔗 Usage Examples + +**Basic Usage**: ```bash -/workflow:execute # Auto-discover active session -/workflow:execute --resume-session="WFS-docs-..." # Specify session -/task:execute IMPL-001 # Execute individual task +# Generate SKILL package for current project +/memory:skill-memory + +# Specify target directory +/memory:skill-memory /path/to/project + +# Force regeneration with Qwen +/memory:skill-memory --tool qwen --regenerate + +# Partial mode (modules only) +/memory:skill-memory --mode partial + +# CLI execution mode +/memory:skill-memory --cli-execute ``` -**Related commands**: -- `/workflow:status` - View task progress -- `/workflow:session:complete` - Mark session complete +**Output**: +``` +✅ SKILL Package Generation Complete + +Project: my_project +Documentation: .workflow/docs/my_project/ (15 files) +SKILL Index: .claude/skills/my_project/SKILL.md + +Generated: +- 4 documentation tasks completed +- SKILL.md with progressive loading (4 levels) +- Module index with 8 modules + +Usage: +- Load Level 0: Quick project overview (~2K tokens) +- Load Level 1: Core modules (~8K tokens) +- Load Level 2: Complete docs (~25K tokens) +- Load Level 3: Everything (~40K tokens) +``` --- - ## [5.1.0] - 2025-10-27 ### 🔄 Agent Architecture Consolidation