使用说明
代理协议
AI原生问题跟踪器 专为居住在终端中的开发人员设计。
使用依赖关系图、TUI仪表板和MCP服务器集成进行快速、键盘驱动的问题跟踪。
✨ 特性
🎯 核心问题管理
- 创建和跟踪:状态、优先级、工作量估算、标签和依赖关系方面的问题
- 智能工作流:自动状态转换(积压→ 准备→ 进行中→ Done)
- 快捷操作:批量操作、检查点、上下文切换
- 模糊搜索:按部分ID、标题或标签查找问题
📊 可视化和分析
- TUI仪表板:全屏交互界面(Jira风格统一列表)
- 依赖图:问题关系的ASCII艺术可视化
- 关键路径:查找依赖链中的瓶颈
- 指标:轨道完成率、速度和燃尽率
🤖 AI集成
- MCP服务器:通过模型上下文协议公开问题跟踪器
- 代理商友好:结构化数据、JSON输出、语义操作
- 情境感知:关注相关问题,显示被阻止的内容
🔧 开发者体验
- Git集成:自动检测分支上下文,提交引用
- 交互模式:复杂操作的引导向导
- 壳牌完井:Bash、Zsh、Fish支持
- 零配置:开箱即用,存储于
~/.local/share/agentx
🎨 TUI仪表板
交互式仪表板在您的终端中提供类似Jira的体验:
特性
- ✅ 统一列表:所有问题都在一个可滚动视图中
- ✅ 实时搜索:类型
/并立即开始搜索 - ✅ 地位分化:状态之间清晰的视觉分离
- ✅ 优先级图标: 🔴 关键,🟡 高,🟢 中等,⚪ 低
- ✅ 标签显示:查看所有内联标签
- ✅ 工作量估算:跟踪时间承诺
- ✅ Vim密钥绑定:
hjkl导航+搜索
🚀 快速开始
安装
# Clone and build
git clone https://github.com/can1357/agentx.git
cd agentx
cargo build --release
# Install to PATH
cargo install --path .
创建您的第一个问题
# Quick create
agentx new "Fix authentication bug" -p high -e 2d -t bug -t security
# Interactive mode
agentx new -i
启动TUI仪表板
agentx ui
# or use the alias:
agentx dash
导航:
↑↓或jk-浏览问题/-搜索(实时结果)g/G-跳到顶部/底部PgUp/PgDn-滚动速度更快Tab-切换窗格q-退出
📖 命令参考
问题生命周期
# Create new issue
agentx new "Task description" [OPTIONS]
# or use the alias:
agentx add "Task description" [OPTIONS]
-p, --priority <critical|high|medium|low>
-e, --effort <duration> # e.g., "2d", "4h", "1w"
-t, --tag <tag> # Can be used multiple times
-d, --depends <ID> # Add dependency
-i, --interactive # Launch wizard
# Update status
agentx start <ID> # Mark as in-progress
agentx block <ID> <reason> # Mark as blocked
agentx close <ID> # Mark as done
agentx defer <ID> # Move to backlog
agentx activate <ID> # Activate from backlog
# View issues
agentx list # All open issues
# or use the alias:
agentx ls # All open issues
agentx show <ID> # Full details
agentx context # Current work context
agentx focus # Top priorities
agentx blocked # All blocked issues
agentx ready # Ready to start
agentx quick-wins # Low-effort tasks
依赖项
# Add dependency
agentx depend add <ID> <depends-on>
# Remove dependency
agentx depend remove <ID> <depends-on>
# Visualize graph
agentx deps-graph [ID] # ASCII art
agentx dependencies <ID> # Show deps + dependents
# Find bottlenecks
agentx critical-path # Longest dependency chain
标签和组织
# Add tags
agentx tag add <ID> <tag>
# Remove tags
agentx tag remove <ID> <tag>
# List by tag (in list command)
agentx list -t backend -t api
批量操作
# Start multiple issues
agentx bulk-start <ID1> <ID2> <ID3>
# Close multiple issues
agentx bulk-close <ID1> <ID2> <ID3>
分析
# Show metrics
agentx metrics
# Recent changes
agentx summary
# Issue counts by status
agentx list --json | jq '.[] | .metadata.status' | sort | uniq -c
🔌 MCP服务器
AgentX可以作为MCP服务器运行,用于AI助手集成:
# Start server on stdio
agentx serve
可用工具
MCP服务器将这些操作暴露给AI助手:
| 工具 | 说明 |
|---|---|
create_issue |
使用元数据创建新问题 |
list_issues |
按状态/优先级/标签查询问题 |
update_issue |
更改状态、添加标签、更新字段 |
show_issue |
获取具体问题的完整详细信息 |
add_dependency |
将问题联系在一起 |
get_blocked |
查找所有被阻止的问题 |
get_ready |
查找准备开始的问题 |
Claude桌面配置示例
{
"mcpServers": {
"agentx": {
"command": "agentx",
"args": ["serve"],
"env": {}
}
}
}
🛠️ 高级用法
从YAML导入
# Bulk import issues
agentx import issues.yaml
格式:
issues:
- title: "Setup CI/CD pipeline"
priority: high
effort: "1w"
tags: ["devops", "infrastructure"]
depends_on: []
- title: "Write integration tests"
priority: medium
effort: "3d"
tags: ["testing"]
depends_on: ["BUG-123"]
别名管理
# Create short aliases for long IDs
agentx alias add auth BUG-123
agentx show auth # Same as: agentx show BUG-123
# List aliases
agentx alias list
检查点
跟踪单个问题的进度:
# Add checkpoint
agentx checkpoint BUG-123 "Completed database migration"
agentx checkpoint BUG-123 "Updated API endpoints"
# View in issue details
agentx show BUG-123
JSON输出
所有命令都支持JSON输出用于脚本编写:
# Get JSON output
agentx list --json | jq '.[] | select(.metadata.priority == "Critical")'
# Count issues by status
agentx list --json | jq 'group_by(.metadata.status) | map({status: .[0].metadata.status, count: length})'
📁 项目结构
agentx/
├── src/
│ ├── main.rs # CLI entrypoint
│ ├── issue.rs # Core issue types
│ ├── storage.rs # File-based persistence
│ ├── commands/ # CLI commands
│ ├── tui/ # Dashboard UI
│ │ ├── mod.rs # App state & event loop
│ │ ├── widgets/ # Kanban, graphs, charts
│ │ └── views/ # Dashboard layout
│ ├── mcp/ # MCP server
│ └── utils/ # Helpers
├── data/ # Issue storage (auto-created)
├── Cargo.toml
└── README.md
数据位置: ~/.local/share/agentx/
🎯 工作流示例
日常工作流程
# Morning: Check context
agentx focus # What's important today?
agentx context # What am I working on?
# Start working
agentx start BUG-123
agentx checkpoint BUG-123 "Initial investigation complete"
# Found blocker
agentx block BUG-123 "Waiting for DB migration approval"
# Switch to quick win
agentx quick-wins
agentx start DOC-45
# End of day
agentx close DOC-45
agentx summary # What did I accomplish?
规划一个功能
# Create feature breakdown
agentx new "API authentication layer" -p high -e 2w -t api
agentx new "Add JWT token generation" -p high -e 3d -t api -d FEA-100
agentx new "Implement token refresh" -p medium -e 2d -t api -d FEA-101
agentx new "Add rate limiting" -p low -e 1d -t api -d FEA-101
# Visualize dependencies
agentx deps-graph FEA-100
# Find critical path
agentx critical-path
缺陷类选
# Import bugs from file
agentx import bugs.yaml
# Review critical issues
agentx list -p critical
# Launch dashboard for triage
agentx dash
# (In dashboard: use / to search, mark priorities)
🔧 配置
AgentX使用合理的默认值,但可以自定义:
# Initialize config
agentx init
# Location: ~/.config/agentx/config.yaml
config.yaml示例:
# Issue ID format
id_prefix: "TASK"
id_start: 1
# Default values
defaults:
priority: medium
effort: null
tags: []
# TUI theme
theme:
name: "default"
🧪 测试
# Run tests
cargo test
# Run with logging
RUST_LOG=debug agentx dash
# Check performance
hyperfine 'agentx list'
📜 许可证
MIT许可证-有关详细信息,请参阅许可证文件。