astercloud · wutongci · Dec 4, 2025 · Dec 4, 2025
diff --git a/.githooks/pre-commit b/.githooks/pre-commit
diff --git a/.github/workflows/go-ci.yml b/.github/workflows/go-ci.yml
@@ -40,7 +40,7 @@ jobs:
       - name: Setup Go
         uses: actions/setup-go@v5
         with:
-          go-version: '1.24'
+          go-version: "1.24"
           cache: true
 
       - name: golangci-lint
@@ -57,7 +57,7 @@ jobs:
     timeout-minutes: 10
     strategy:
       matrix:
-        go-version: ['1.24']
+        go-version: ["1.24"]
     steps:
       - name: Checkout code
         uses: actions/checkout@v4
@@ -94,7 +94,7 @@ jobs:
       - name: Setup Go
         uses: actions/setup-go@v5
         with:
-          go-version: '1.24'
+          go-version: "1.24"
           cache: true
 
       - name: Run tests

diff --git a/.gitignore b/.gitignore
@@ -207,3 +207,8 @@ article-writer-web-service/
 # claude
 .claude/
 *_PLAN.md
+
+# lwmacct conventions
+.local/
+.taskfile/
+Taskfile.yml
diff --git a/.golangci.yml b/.golangci.yml
@@ -4,7 +4,7 @@ version: "2"
 run:
   timeout: 5m
   tests: true
-  go: '1.21'
+  go: "1.21"
 
 linters:
   enable:

diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
@@ -14,6 +14,7 @@ repos:
       - id: trailing-whitespace
       - id: end-of-file-fixer
       - id: check-yaml
+        args: [--allow-multiple-documents]
       - id: check-added-large-files
-        args: ['--maxkb=1000']
+        args: ["--maxkb=1000"]
       - id: check-merge-conflict
diff --git a/.prettierrc b/.prettierrc
@@ -0,0 +1,4 @@
+{
+  "tabWidth": 2,
+  "printWidth": 256
+}
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -3,25 +3,30 @@
 ## [v0.15.0] - 2025-11-24
 
 ### Added
+
 - 完成15个工具的全面测试验证
 - 添加工具测试文档和报告
 
 ### Fixed
+
 - **WebSearch工具**: 修复search_depth参数问题，从"general"改为"basic"以符合Tavily API规范
 - 验证多轮推理功能正常工作（streaming.go）
 
 ### Tested
+
 - 网络工具: HttpRequest ✅, WebSearch ✅
 - 文件工具: Read ✅, Write ✅, Edit ✅, Glob ✅, Grep ✅
 - 执行工具: Bash ✅, BashOutput ✅, KillShell ✅
 - 任务管理: TodoWrite ✅, Task ✅, ExitPlanMode ✅
 - 高级工具: DemoLongTask ✅, Skill ⚠️, SemanticSearch ⚠️
 
 ### Notes
+
 - Skill和SemanticSearch工具需要额外配置才能完整使用
 - 所有核心工具调用功能已验证正常
 - 系统可投入实际使用
 
 ## [v0.14.0] - Previous Release
+
 - 基础工具实现
 - 多轮推理支持
diff --git a/CLAUDE.md b/CLAUDE.md
@@ -68,44 +68,37 @@ Aster is an event-driven AI Agent framework built with Go, implementing the Goog
 ### Key Components
 
 1. **Agent System** (`pkg/agent/`)
-
    - Core agent implementation with event-driven architecture
    - Three event channels: Progress (real-time output), Control (human interaction), Monitor (governance)
    - Dependencies injection pattern with Registry pattern for tools and templates
 
 2. **Workflow Engine** (`pkg/workflow/`)
-
    - Sequential, Parallel, and Loop workflow agents
    - 8-step types with Router for dynamic routing
    - Stream-based execution with Go 1.23 iter.Seq2
 
 3. **Memory Management** (`pkg/memory/`)
-
    - Three-tier system: Text Memory, Working Memory, Semantic Memory
    - Advanced features: Provenance (source tracking), Consolidation (LLM-driven merging), PII Auto-Redaction
    - Vector store integration with confidence scoring
 
 4. **Middleware System** (`pkg/middleware/`)
-
    - Onion-model architecture with priority-based layers
    - Built-in middlewares: filesystem, summarization, agent memory, working memory
    - Custom middleware support via WrapModelCall/WrapToolCall interfaces
 
 5. **Tools System** (`pkg/tools/`)
-
    - Registry pattern for tool discovery and management
    - Built-in tools: filesystem operations, bash execution, HTTP requests, web search, todo management
    - MCP (Model Context Protocol) support for external tool integration
    - Long-running tools with async execution and progress tracking
 
 6. **Session & Persistence** (`pkg/session/`)
-
    - PostgreSQL and MySQL support with JSONB/JSON columns
    - Event sourcing with append-only storage
    - 7-point recovery mechanism for fault tolerance
 
 7. **Multi-Agent Collaboration** (`pkg/stars/`)
-
    - Stars pattern for agent collaboration
    - Scheduler for intelligent task distribution
    - Permission management system

diff --git a/README_EN.md b/README_EN.md
@@ -21,6 +21,7 @@
 **Aster** (星尘云枢) is a production-ready AI Agent development framework built with Go, designed to run agents safely and efficiently in enterprise environments.
 
 Like stardust converging to form a celestial hub, Aster brings together:
+
 - **High Performance**: Go's concurrency model supports 100+ concurrent agents
 - **Native Script Execution**: Run Python, TypeScript, and Bash natively with full sandbox isolation
 - **Event-Driven Architecture**: Progress/Control/Monitor tri-channel design for clear separation of concerns
@@ -29,33 +30,39 @@ Like stardust converging to form a celestial hub, Aster brings together:
 ## 🎯 Core Features
 
 ### 🚀 Multi-Language Script Execution
+
 - **Python**: Execute data processing, ML workflows, and analytics scripts
-- **TypeScript**: Run modern JavaScript/TypeScript for web automation and API interactions  
+- **TypeScript**: Run modern JavaScript/TypeScript for web automation and API interactions
 - **Bash**: Shell commands for system operations and DevOps tasks
 - All with **native performance** and **isolated sandbox environments**
 
 ### 🎪 Event-Driven Architecture
+
 - **Progress Channel**: Real-time text streaming, tool execution progress
 - **Control Channel**: Tool approval requests, human-in-the-loop interactions
 - **Monitor Channel**: Governance events, error tracking, audit logs
 
 ### 🧅 Middleware Onion Model
+
 - **Layered Processing**: Requests and responses flow through middleware layers
 - **Built-in Middlewares**: Auto-summarization, PII redaction, tool interception
 - **Extensible**: Create custom middleware for your specific needs
 
 ### 🔒 Enterprise-Grade Security
+
 - **Cloud Sandbox**: Native integration with Aliyun AgentBay, Volcengine
 - **PII Auto-Redaction**: Detect and redact 10+ types of sensitive data
 - **Permission System**: Fine-grained tool-level access control
 - **Audit Logging**: Complete tool call tracking and state management
 
 ### 🧠 Three-Layer Memory System
+
 - **Text Memory**: File-based short-term memory for conversation context
 - **Working Memory**: Persistent state across sessions with TTL and JSON schema
 - **Semantic Memory**: Vector-based long-term memory with provenance tracking
 
 ### 🔄 Advanced Capabilities
+
 - **Streaming API**: iter.Seq2-based streaming with 80%+ memory reduction
 - **Long-Running Tools**: Async task management with progress tracking
 - **Multi-Agent Orchestration**: Pool/Room/Workflow collaboration patterns
@@ -91,7 +98,7 @@ func main() {
     // 1. Setup dependencies
     toolRegistry := tools.NewRegistry()
     builtin.RegisterAll(toolRegistry)
-    
+
     jsonStore, _ := store.NewJSONStore("./.aster")
     deps := &agent.Dependencies{
         Store:            jsonStore,
@@ -162,13 +169,15 @@ func main() {
 ![Middleware Onion](https://raw.githubusercontent.com/astercloud/aster/main/docs/public/images/middleware-onion.svg)
 
 The middleware architecture processes each request/response through multiple layers:
+
 - Higher priority middleware sits in outer layers
 - Handles requests first, processes responses last
 - Clean separation of concerns and easy extensibility
 
 ## 🌐 Multi-Language Execution
 
 ### Python Script Example
+
 ```python
 # agent can execute this directly
 import pandas as pd
@@ -179,18 +188,20 @@ print(result.to_json())
 ```
 
 ### TypeScript Example
+
 ```typescript
 // Native TypeScript execution
 interface User {
   name: string;
   email: string;
 }
 
-const users: User[] = await fetch('/api/users').then(r => r.json());
-console.log(users.map(u => u.name));
+const users: User[] = await fetch("/api/users").then((r) => r.json());
+console.log(users.map((u) => u.name));
 ```
 
 ### Bash Example
+
 ```bash
 # System operations
 find . -name "*.log" -mtime +7 -delete
@@ -203,15 +214,16 @@ docker ps | grep running
 
 ### Completed Phases
 
-✅ **Phase 1**: Foundation (Event system, Sandbox abstraction, Storage)  
-✅ **Phase 2**: Agent Runtime (Message processing, Tool system, Streaming)  
-✅ **Phase 3**: Cloud Integration (MCP, Aliyun, Volcengine)  
-✅ **Phase 4**: Multi-Agent (Pool, Room, Scheduler, Permissions)  
-✅ **Phase 5**: MCP Support (Protocol, Servers, Tools)  
-✅ **Phase 6**: Advanced Features (Commands, Skills, Middleware, Multi-provider)  
+✅ **Phase 1**: Foundation (Event system, Sandbox abstraction, Storage)
+✅ **Phase 2**: Agent Runtime (Message processing, Tool system, Streaming)
+✅ **Phase 3**: Cloud Integration (MCP, Aliyun, Volcengine)
+✅ **Phase 4**: Multi-Agent (Pool, Room, Scheduler, Permissions)
+✅ **Phase 5**: MCP Support (Protocol, Servers, Tools)
+✅ **Phase 6**: Advanced Features (Commands, Skills, Middleware, Multi-provider)
 ✅ **Phase 7**: ADK Alignment (Streaming, OpenTelemetry, Persistence, Workflows)
 
 **Current Stats**:
+
 - ~18,000+ LOC
 - 25+ new modules
 - 80%+ test coverage
@@ -221,16 +233,16 @@ docker ps | grep running
 
 Aster fully implements the **Google Context Engineering** whitepaper's 8 core capabilities:
 
-| Capability | Status | Description |
-|------------|--------|-------------|
-| Sessions & Memory | ✅ | Three-layer memory system (Text/Working/Semantic) |
-| Memory Provenance | ✅ | Source tracking with confidence scoring |
-| Memory Consolidation | ✅ | LLM-driven intelligent memory merging |
-| PII Auto-Redaction | ✅ | Automated privacy data protection |
-| Event-Driven Architecture | ✅ | Progress/Control/Monitor tri-channel |
-| Streaming & Backpressure | ✅ | iter.Seq2 streaming interface |
-| Multi-Agent Orchestration | ✅ | Pool/Room/Workflow patterns |
-| Observability | ✅ | Complete OpenTelemetry integration |
+| Capability                | Status | Description                                       |
+| ------------------------- | ------ | ------------------------------------------------- |
+| Sessions & Memory         | ✅     | Three-layer memory system (Text/Working/Semantic) |
+| Memory Provenance         | ✅     | Source tracking with confidence scoring           |
+| Memory Consolidation      | ✅     | LLM-driven intelligent memory merging             |
+| PII Auto-Redaction        | ✅     | Automated privacy data protection                 |
+| Event-Driven Architecture | ✅     | Progress/Control/Monitor tri-channel              |
+| Streaming & Backpressure  | ✅     | iter.Seq2 streaming interface                     |
+| Multi-Agent Orchestration | ✅     | Pool/Room/Workflow patterns                       |
+| Observability             | ✅     | Complete OpenTelemetry integration                |
 
 **100% Implementation** - First Go framework to fully implement the standard.
 
@@ -239,11 +251,13 @@ Aster fully implements the **Google Context Engineering** whitepaper's 8 core ca
 Aster builds upon the excellent work of the open-source community:
 
 ### Frameworks
+
 - **[LangChain](https://github.com/langchain-ai/langchain)**: Pioneering agent framework
 - **[Google ADK](https://github.com/google/genkit)**: Enterprise-grade agent toolkit
 - **[Claude Agent SDK](https://github.com/anthropics/anthropic-sdk-python)**: Computer Use & MCP reference
 
 ### Research
+
 Special thanks to the **[Google Context Engineering Whitepaper](https://cloud.google.com/blog/products/ai-machine-learning/context-engineering-for-ai-agents)** for defining agent capabilities and best practices.
 
 ## 📄 License