using-agentops

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

AgentOps Workflow

AgentOps工作流

You have access to the AgentOps skill set for structured development workflows.

你可以使用AgentOps技能集来进行结构化的开发工作流。

The RPI Workflow

RPI工作流

Research → Plan → Implement → Validate
    ↑                            │
    └──── Knowledge Flywheel ────┘

Research → Plan → Implement → Validate
    ↑                            │
    └──── Knowledge Flywheel ────┘

Research Phase

研究阶段

bash

/research <topic>      # Deep codebase exploration
/knowledge <query>     # Query existing knowledge

Output:

.agents/research/<topic>.md

bash

/research <topic>      # 深度代码库探索
/knowledge <query>     # 查询现有知识

输出：

.agents/research/<topic>.md

Plan Phase

规划阶段

bash

/pre-mortem <spec>     # Simulate failures before implementing
/plan <goal>           # Decompose into trackable issues

Output: Beads issues with dependencies

bash

/pre-mortem <spec>     # 在实现前模拟故障
/plan <goal>           # 分解为可追踪的任务

输出： 带有依赖关系的Beads任务

Implement Phase

实现阶段

bash

/implement <issue>     # Single issue execution
/crank <epic>          # Autonomous epic loop (uses swarm for waves)
/swarm                 # Parallel execution (fresh context per agent)

Output: Code changes, tests, documentation

bash

/implement <issue>     # 执行单个任务
/crank <epic>          # 自主史诗级任务循环（使用swarm进行多轮执行）
/swarm                 # 并行执行（每个Agent拥有独立上下文）

输出： 代码变更、测试、文档

Validate Phase

验证阶段

bash

/vibe [target]         # Code validation (security, quality, architecture)
/post-mortem           # Extract learnings after completion
/retro                 # Quick retrospective

Output:

.agents/learnings/

.agents/patterns/

bash

/vibe [target]         # 代码验证（安全性、质量、架构）
/post-mortem           # 完成后提取经验
/retro                 # 快速回顾

输出：

.agents/learnings/

.agents/patterns/

Phase-to-Skill Mapping

阶段与技能映射表

Phase	Primary Skill	Supporting Skills
Research	`/research`	`/knowledge` , `/inject`
Plan	`/plan`	`/pre-mortem`
Implement	`/implement`	`/crank` (epic loop), `/swarm` (parallel execution)
Validate	`/vibe`	`/retro` , `/post-mortem`

Choosing the skill:

Use
```
/implement
```
for single issue execution.
Use
```
/crank
```
for autonomous epic execution (loops waves via swarm until done).
Use
```
/swarm
```
directly for parallel execution without beads (TaskList only).
Use
```
/ratchet
```
to gate/record progress through RPI.

阶段	核心技能	支持技能
研究	`/research`	`/knowledge` , `/inject`
规划	`/plan`	`/pre-mortem`
实现	`/implement`	`/crank` （史诗级任务循环）, `/swarm` （并行执行）
验证	`/vibe`	`/retro` , `/post-mortem`

技能选择指南：

针对单个任务执行，使用
```
/implement
```
。
针对自主史诗级任务执行，使用
```
/crank
```
（通过swarm循环多轮直至完成）。
若无需Beads（仅使用任务列表）进行并行执行，直接使用
```
/swarm
```
。
使用
```
/ratchet
```
来管控/记录RPI流程中的进度。

Available Skills

可用技能

Skill	Purpose
`/research`	Deep codebase exploration
`/pre-mortem`	Failure simulation before implementing
`/plan`	Epic decomposition into issues
`/implement`	Execute single issue
`/crank`	Autonomous epic loop (uses swarm for each wave)
`/swarm`	Fresh-context parallel execution (Ralph pattern)
`/vibe`	Code validation
`/retro`	Extract learnings
`/post-mortem`	Full validation + knowledge extraction
`/beads`	Issue tracking operations
`/bug-hunt`	Root cause analysis
`/knowledge`	Query knowledge artifacts
`/complexity`	Code complexity analysis
`/doc`	Documentation generation
`/provenance`	Trace artifact lineage to sources
`/trace`	Trace design decisions through history

技能	用途
`/research`	深度代码库探索
`/pre-mortem`	实现前故障模拟
`/plan`	将史诗级任务分解为子任务
`/implement`	执行单个任务
`/crank`	自主史诗级任务循环（每轮使用swarm）
`/swarm`	独立上下文并行执行（Ralph模式）
`/vibe`	代码验证
`/retro`	提取经验
`/post-mortem`	完整验证 + 知识提取
`/beads`	任务追踪操作
`/bug-hunt`	根因分析
`/knowledge`	查询知识工件
`/complexity`	代码复杂度分析
`/doc`	文档生成
`/provenance`	追踪工件溯源
`/trace`	追溯历史中的设计决策

Knowledge Flywheel

知识飞轮

Every

/post-mortem

feeds back to

/research

Learnings extracted →
```
.agents/learnings/
```
Patterns discovered →
```
.agents/patterns/
```
Research enriched → Future sessions benefit

每次

/post-mortem

的结果都会反馈给

/research

：

提取的经验 →
```
.agents/learnings/
```
发现的模式 →
```
.agents/patterns/
```
丰富的研究内容 → 未来会话可从中受益

Natural Language Triggers

自然语言触发

Skills auto-trigger from conversation:

Say This	Runs
"I need to understand how auth works"	`/research`
"Check my code for issues"	`/vibe`
"What could go wrong with this?"	`/pre-mortem`
"Let's execute this epic"	`/crank`
"Spawn agents to work in parallel"	`/swarm`
"How did we decide on this?"	`/trace`
"Where did this learning come from?"	`/provenance`

技能会从对话中自动触发：

你可以说	执行的技能
"我需要理解认证机制的工作原理"	`/research`
"检查我的代码是否存在问题"	`/vibe`
"这个方案可能会出什么问题？"	`/pre-mortem`
"我们来执行这个史诗级任务"	`/crank`
"生成多个Agent并行工作"	`/swarm`
"我们当初是如何做出这个决定的？"	`/trace`
"这个经验来自哪里？"	`/provenance`

Issue Tracking

问题追踪

AgentOps uses beads for git-native issue tracking:

bash

bd ready              # Unblocked issues
bd show <id>          # Issue details
bd close <id>         # Close issue
bd sync               # Sync with git

AgentOps使用beads进行原生Git问题追踪：

bash

bd ready              # 查看未阻塞的任务
bd show <id>          # 查看任务详情
bd close <id>         # 关闭任务
bd sync               # 与Git同步