Back to Details

prath-mode

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Prath mode

Prath模式

Read the matched leaf skill in full before executing. Do not restate or improvise its workflow. The leaf owns its own triggers, flags, and rules.
执行前完整阅读匹配到的叶子skill。不要重述或随意修改其工作流。叶子skill拥有自己的触发条件、标记和规则。

Routing map

路由映射表

ActionLeaf skill
Commit or write a commit messagecommit (
../commit/SKILL.md
)
Strip AI slop or simplify a diffdeslop (
../deslop/SKILL.md
)
Open a pull requestmake-pr (
../make-pr/SKILL.md
)
Address PR review feedbackfix-pr (
../fix-pr/SKILL.md
)
Fix or implement a Linear ticketfix-linear-ticket (
../fix-linear-ticket/SKILL.md
)
Review a plan, design, or implementationpeer-review (
../peer-review/SKILL.md
)
Clone or search an external git repobox (
../box/SKILL.md
)
Delegate work to an external agentassign (
../assign/SKILL.md
)
Save session context or resume a handoffhandoff (
../handoff/SKILL.md
)
Explain a diff, branch, or PR (HTML)explain-diff (
../explain-diff/SKILL.md
)
Send a Discord notification or alertnotify (
../notify/SKILL.md
)
When a request matches several skills, pick the leaf that owns the immediate action. For multi-step work, use a chain below.
操作叶子skill
提交或编写提交信息commit (
../commit/SKILL.md
)
去除AI冗余内容或简化差异deslop (
../deslop/SKILL.md
)
发起拉取请求make-pr (
../make-pr/SKILL.md
)
处理PR评审反馈fix-pr (
../fix-pr/SKILL.md
)
修复或实现Linear工单fix-linear-ticket (
../fix-linear-ticket/SKILL.md
)
评审方案、设计或实现peer-review (
../peer-review/SKILL.md
)
克隆或搜索外部Git仓库box (
../box/SKILL.md
)
将工作委派给外部Agentassign (
../assign/SKILL.md
)
保存会话上下文或恢复交接handoff (
../handoff/SKILL.md
)
解释差异、分支或PR（HTML格式）explain-diff (
../explain-diff/SKILL.md
)
发送Discord通知或警报notify (
../notify/SKILL.md
)
当请求匹配多个skill时，选择对应直接操作的叶子skill。对于多步骤工作，使用下方的工作流链。

Workflow chains

工作流链

Run in order. Read each leaf before its step. 
implement
is not a skill.
  1. Ship ticket work: 
    fix-linear-ticket
    → implement → 
    deslop
    → 
    commit
    → 
    make-pr
    (pass 
    --ticket
    to make-pr when the ID is known)
  2. Ship planned work: 
    peer-review
    → implement → 
    deslop
    → 
    commit
    → 
    make-pr
  3. Quick save: 
    deslop
    (optional) → 
    commit
  4. Land PR feedback: 
    fix-pr
    → 
    commit
    → push
  5. Research external code: 
    box
  6. Delegate heavy lift: 
    assign
  7. End or resume session: 
    handoff
    or 
    handoff --resume <path>
按顺序运行。执行每个步骤前先阅读对应的叶子skill。
implement
不是一个skill。
  1. 交付工单工作: 
    fix-linear-ticket
    → implement → 
    deslop
    → 
    commit
    → 
    make-pr
    （当工单ID已知时，向make-pr传递
    --ticket
    参数）
  2. 交付规划工作: 
    peer-review
    → implement → 
    deslop
    → 
    commit
    → 
    make-pr
  3. 快速保存: 
    deslop
    （可选）→ 
    commit
  4. 处理PR反馈: 
    fix-pr
    → 
    commit
    → push
  5. 研究外部代码: 
    box
  6. 委派繁重工作: 
    assign
  7. 结束或恢复会话: 
    handoff
    或 
    handoff --resume <path>

Step 1: Match

步骤1：匹配

Match the request to one leaf or one chain. If nothing matches, say so and ask what the user wants.
将请求与某个叶子skill或某条工作流链匹配。如果没有匹配项，请告知用户并询问其需求。

Step 2: Verify installation

步骤2：验证安装

Before reading or executing a leaf, confirm it is installed: resolve 
../<skill-name>/SKILL.md
relative to this skill's directory and check it exists (
test -f
or the Read tool). For a chain, verify every leaf before the first step and re-check before each subsequent step, the user may install skills between steps.
If any leaf is missing, stop and report its name and checked path, list all missing names in one message, and tell the user to install: 
npx skills@latest add prathamdby/skills
. Never improvise a missing leaf's workflow.
在读取或执行叶子skill前，确认其已安装：相对于当前skill的目录解析
../<skill-name>/SKILL.md
路径，并检查文件是否存在（使用
test -f
或读取工具）。对于工作流链，在第一步前验证所有叶子skill，并且在后续每个步骤前重新检查，因为用户可能在步骤之间安装skill。
如果任何叶子skill缺失，请停止操作并报告其名称和检查路径，在一条消息中列出所有缺失的名称，并告知用户执行以下命令安装： 
npx skills@latest add prathamdby/skills
。绝不要随意修改缺失叶子skill的工作流。

Step 3: Read and invoke

步骤3：读取并调用

Read the matched leaf's 
SKILL.md
in full, then execute per the leaf. For chains, complete each step before starting the next unless the user narrows scope.
完整读取匹配到的叶子skill的
SKILL.md
文件，然后按照该叶子skill的要求执行。对于工作流链，完成每个步骤后再开始下一个步骤，除非用户缩小了范围。