claw-code 源码分析：OmX `$team` / `$ralph`——把 AI 辅助开发从偶发灵感变成可重复流水线

说明OmXoh-my-codex、$team、$ralph在 claw-code 中主要出现在README 叙事与资源目录是外部编排工具的概念名仓库并未实现 OmX 本体。本文说明官方如何描述它们以及同一哲学在源码与仓库制度里如何落成可重复、可验证的流水线。1. 官方定义OmX 与两种模式在 README 中的角色维护者将 Python 移植会话描述为在oh-my-codex (OmX)上端到端编排并点名两种工作模式61:65:README.md The whole thing was orchestrated end-to-end using [oh-my-codex (OmX)](https://github.com/Yeachan-Heo/oh-my-codex) by [bellman_ych](https://x.com/bellman_ych) — a workflow layer built on top of OpenAIs Codex ([OpenAIDevs](https://x.com/OpenAIDevs)). I used $team mode for parallel code review and $ralph mode for persistent execution loops with architect-level verification. The entire porting session — from reading the original harness structure to producing a working Python tree with tests — was driven through OmX orchestration. The result is a clean-room Python rewrite that captures the architectural patterns of Claw Codes agent harness without copying any proprietary source. Im now actively collaborating with [bellman_ych](https://x.com/bellman_ych) — the creator of OmX himself — to push this further. The basic Python foundation is already in place and functional, but were just getting started. **Stay tuned — a much more capable version is on the way.** The Rust port was developed with both [oh-my-codex (OmX)](https://github.com/Yeachan-Heo/oh-my-codex) and [oh-my-opencode (OmO)](https://github.com/code-yeongyu/oh-my-openagent): OmX drove scaffolding, orchestration, and architecture direction, while OmO was used for later implementation acceleration and verification support.文末再次归纳模式语义与洁净室、人工校验并列194:199:README.md Key workflow patterns used during the port: - **$team mode:** coordinated parallel review and architectural feedback - **$ralph mode:** persistent execution, verification, and completion discipline - **Cleanroom passes:** naming/branding cleanup, QA, and release validation across the Rust workspace - **Manual and live validation:** build, test, manual QA, and real API-path verification before publish读法$team强调并行、评审、架构反馈$ralph强调持久执行环、验证、完工纪律。二者都不是 claw-code 里的函数名而是人机协作节奏的标签可重复性来自「节奏 验收标准」而不是单次神提示词。2. 仓库如何把「节奏」固化成流水线制品2.1 验收门禁unittest discover与 CLI 子命令网README 把验证路径写进 Quickstart与 OmX 叙事同一文档内并列——谁 clone 都能复跑Run verification: bash python3 -m unittest discover -s tests -v这与WorkspaceSetup.test_commandsrc/setup.py一致构成与 IDE/Agent 无关的硬门禁对应分析见result/26.md。$ralph的工程化对应物不是「模型多聊几句」而是红绿可判定的checkTrue子进程、parity-audit、bootstrap/turn-loop 等可脚本化步骤。2.2 并行评审与架构反馈$team在仓库内的「影子」README 用assets/omx/截图展示分屏评审caption 中直接出现 Ralph/team orchestration### OmX workflow screenshots  *Ralph/team orchestration view while the README and essay context were being reviewed in terminal panes.*  *Split-pane review and verification flow during the final README wording pass.*在代码侧「多视角对齐」更接近parity-auditreference_data/*把「像不像」从主观变成比例与缺失列表src/parity_audit.py。PARITY.mdRust/TS 表面只读对比的长文评审记录显式写not feature-parity避免口头并行评审后仍假装完成。这些不依赖 OmX 运行但承担$team所追求的「架构反馈可被引用」之功能。2.3 编排状态不进库.omx忽略与 diff 排除根目录与rust/.gitignore均忽略.omx/暗示OmX 本地状态与可发布源码分离__pycache__/ archive/ .omx/ .clawd-agents/claw的 diff 报告对 git diff 排除.omx避免终端工作流垃圾污染「工作区是否干净」的判定// 2267:2270:rust/crates/claw-cli/src/main.rsfnrender_diff_report()-ResultString,Boxdynstd::error::Error{letoutputstd::process::Command::new(git).args([diff,--,:(exclude).omx]).current_dir(env::current_dir()?)流水线哲学易变、机密的编排缓存留在本机契约与快照tests/、reference_data/、文档进 Git——重复跑的是后者。3. 与产品内「多角色」的弱类比非同一实现RustAgent工具按subagent_type划分工具白名单例如Explore / Plan / Verification探索、规划、带 bash 的验证与$team多视角/$ralph执行—验收闭环在隐喻上同构但是运行时内置策略不是 OmX// 1665:1701:rust/crates/tools/src/lib.rsfnallowed_tools_for_subagent(subagent_type:str)-VecString{lettoolsmatchsubagent_type{Explorevec![read_file,glob_search,...],Planvec![...TodoWrite,...],Verificationvec![bash,read_file,...],区分清楚OmX 是开发 claw-code 时用的外壳Subagent 类型是 claw-code 交付给用户的能力。二者都体现「别把智能体当成单轮聊天」但层级不同。4. 小结从偶发灵感到流水线的翻译表OmX 叙事中的概念在 claw-code 仓库中的可触摸对应OmX 编排层外部工具仓库用 README 截图 .gitignore记录其角色不内嵌实现$team并行评审与架构反馈 →PARITY.md、parity-audit、reference_data、多子系统 manifest$ralph持久执行 验证纪律 →unittest discover、大量src.mainCLI 黑盒测、bootstrap/turn-loop链、cargo 构建与测试及 README 中的 cleanroom / manual validationOmORust 阶段README 所述加速实现与验证支持与 Python 阶段分工并列可重复标准命令 版本化快照 条件 parity无 archive 不谎称对齐见result/26.md一句话$team/$ralph是工作流层的命名claw-code 用测试、审计 CLI、快照与诚实 parity 文案把同一意图落成任何人都能跑的流水线——AI 辅助从「某次对话很顺」变成「门禁与文档可继承」。