gm-execute

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

GM EXECUTE — Resolving Every Unknown

GM EXECUTE — 解析所有未知项

You are in the EXECUTE phase. Resolve every named mutable via witnessed execution. Any new unknown = stop, snake to

planning

, restart chain.

GRAPH POSITION:

PLAN → [EXECUTE] → EMIT → VERIFY → COMPLETE

Entry: .prd exists with all unknowns named. Entered from
```
planning
```
or via snake from EMIT/VERIFY.

你正处于EXECUTE执行阶段。通过可观测的执行解析所有已命名的可变项。若出现任何新的未知项，立即停止，回退至

planning

规划阶段，重启流程链。

流程节点位置：

PLAN → [EXECUTE] → EMIT → VERIFY → COMPLETE

进入条件：存在已命名所有未知项的.prd文件。可从
```
planning
```
规划阶段进入，或从EMIT/VERIFY阶段回退进入。

TRANSITIONS

流转规则

FORWARD: All mutables KNOWN → invoke

gm-emit

skill

SELF-LOOP: Mutable still UNKNOWN after one pass → re-run with different angle (max 2 passes then snake)

BACKWARD:

New unknown discovered → invoke
```
planning
```
skill immediately, restart chain
From EMIT: logic error → re-enter here, re-resolve mutable
From VERIFY: runtime failure → re-enter here, re-resolve with real system state

向前流转：所有可变项均已明确 → 调用

gm-emit

技能

自循环：一次执行后仍有可变项未知 → 换角度重新执行（最多2次，之后回退至规划阶段）

向后回退:

发现新的未知项 → 立即调用
```
planning
```
技能，重启流程链
从EMIT阶段返回：逻辑错误 → 重新进入本阶段，重新解析可变项
从VERIFY阶段返回：运行时失败 → 重新进入本阶段，结合真实系统状态重新解析

MUTABLE DISCIPLINE

可变项规范

Each mutable: name | expected | current | resolution method. Execute → witness → assign → compare. Zero variance = resolved. Unresolved after 2 passes = new unknown = snake to

planning

. Never narrate past an unresolved mutable.

每个可变项需包含：名称 | 预期值 | 当前值 | 解析方法。执行→观测→赋值→对比。无差异即表示已解析。若2次执行后仍未解析，则视为新的未知项，回退至

planning

规划阶段。不得跳过未解析的可变项进行后续描述。

CODE EXECUTION

代码执行

exec:<lang> is the only way to run code. Bash tool body:

exec:<lang>\n<code>

exec:nodejs

(default) |

exec:bash

exec:python

exec:typescript

exec:go

exec:rust

exec:c

exec:cpp

exec:java

exec:deno

exec:cmd

Lang auto-detected if omitted.

cwd

sets directory. File I/O via exec:nodejs + require('fs'). Only git in bash directly.

Bash(node/npm/npx/bun)

= violations.

Background tasks (auto-backgrounded when execution exceeds 15s):

exec:sleep
<task_id> [seconds]

exec:status
<task_id>

exec:close
<task_id>

Runner (PM2-backed — all activity visible in

pm2 list

and

pm2 monit

in user terminal):

exec:runner
start|stop|status

exec:<lang>是运行代码的唯一方式。Bash工具格式：

exec:<lang>\n<code>

exec:nodejs

（默认）|

exec:bash

exec:python

exec:typescript

exec:go

exec:rust

exec:c

exec:cpp

exec:java

exec:deno

exec:cmd

若省略语言类型，将自动检测。

cwd

用于设置工作目录。文件I/O需通过exec:nodejs + require('fs')实现。仅可在bash中直接使用git命令。

Bash(node/npm/npx/bun)

属于违规操作。

后台任务（当执行时间超过15秒时自动转入后台）：

exec:sleep
<task_id> [seconds]

exec:status
<task_id>

exec:close
<task_id>

运行器（基于PM2实现——所有活动可在用户终端的

pm2 list

和

pm2 monit

中查看）：

exec:runner
start|stop|status

CODEBASE EXPLORATION

代码库探索

exec:codesearch
<natural language description of what you need>

Alias:

exec:search

. Glob, Grep, Read-for-discovery, Explore, WebSearch = blocked.

exec:codesearch
<你所需内容的自然语言描述>

别名：

exec:search

。Glob、Grep、Read-for-discovery、Explore、WebSearch均被禁用。

IMPORT-BASED DEBUGGING

基于导入的调试

Always import actual codebase modules. Never rewrite logic inline.

exec:nodejs
const { fn } = await import('/abs/path/to/module.js');
console.log(await fn(realInput));

Witnessed import output = resolved mutable. Reimplemented output = UNKNOWN.

始终导入真实的代码库模块。切勿内重写逻辑。

exec:nodejs
const { fn } = await import('/abs/path/to/module.js');
console.log(await fn(realInput));

观测到的导入输出即表示可变项已解析。重实现的输出视为未知。

EXECUTION DENSITY

执行密度

Pack every related hypothesis into one run. Each run ≤15s. Witnessed output = ground truth. Narrated assumption = nothing.

Parallel waves: ≤3

gm:gm

subagents via Task tool — independent items simultaneously, never sequentially.

将所有相关假设整合到一次执行中。每次执行时长≤15秒。观测到的输出即为真实结果。口头假设无效。

并行执行：通过Task工具最多可启动3个

gm:gm

子代理——同时处理独立项，切勿按顺序执行。

CHAIN DECOMPOSITION

流程链分解

Break every multi-step operation before running end-to-end:

Number every distinct step
Per step: input shape, output shape, success condition, failure mode
Run each step in isolation — witness — assign mutable — KNOWN before next
Debug adjacent pairs for handoff correctness
Only when all pairs pass: run full chain end-to-end

Step failure revealing new unknown → snake to

planning

在端到端运行前，拆分所有多步骤操作：

为每个不同步骤编号
每个步骤需明确：输入格式、输出格式、成功条件、失败模式
独立运行每个步骤——观测——赋值可变项——确认已明确后再进行下一步
调试相邻步骤对，确保交接正确性
仅当所有步骤对均通过后，再端到端运行完整流程链

若步骤失败并发现新的未知项，回退至

planning

规划阶段。

BROWSER DEBUGGING

浏览器调试

Invoke

agent-browser

skill. Escalation — exhaust each before advancing:

```
exec:agent-browser\n<js>
```
— query DOM/state. Always first.
```
agent-browser
```
skill +
```
__gm
```
globals — instrument and capture
navigate/click/type — only when real events required
screenshot — last resort

__gm

scaffold:

window.__gm = { captures: [], log: (...a) => window.__gm.captures.push({t:Date.now(),a}), assert: (l,c) => { window.__gm.captures.push({l,pass:!!c,val:c}); return !!c; }, dump: () => JSON.stringify(window.__gm.captures,null,2) };

调用

agent-browser

技能。按优先级逐步升级——在进入下一环节前需穷尽当前方式：

```
exec:agent-browser\n<js>
```
— 查询DOM/状态。始终作为首选方式。
```
agent-browser
```
技能 +
```
__gm
```
全局变量 — 插桩并捕获数据
导航/点击/输入 — 仅当需要真实事件时使用
截图 — 最后手段

__gm

脚手架：

window.__gm = { captures: [], log: (...a) => window.__gm.captures.push({t:Date.now(),a}), assert: (l,c) => { window.__gm.captures.push({l,pass:!!c,val:c}); return !!c; }, dump: () => JSON.stringify(window.__gm.captures,null,2) };

GROUND TRUTH

真实基准

Real services, real data, real timing. Mocks/fakes/stubs = delete immediately. No .test.js/.spec.js. Delete on discovery.

使用真实服务、真实数据、真实时序。Mocks/fakes/stubs需立即删除。禁止使用.test.js/.spec.js文件，发现后立即删除。

CONSTRAINTS

约束规则

Never:

Bash(node/npm/npx/bun)

Always: witness every hypothesis | import real modules | snake to planning on any new unknown | fix immediately on discovery

→ FORWARD: All mutables KNOWN → invoke

gm-emit

skill. ↺ SELF-LOOP: Still UNKNOWN → re-run (max 2 passes). ↩ SNAKE to PLAN: Any new unknown → invoke

planning

skill, restart chain.

禁止：

Bash(node/npm/npx/bun)

必须：观测每个假设 | 导入真实模块 | 出现任何新未知项时回退至规划阶段 | 发现问题立即修复

→ 向前流转：所有可变项均已明确 → 调用

gm-emit

技能。 ↺ 自循环：仍有可变项未知 → 重新执行（最多2次）。 ↩ 回退至规划阶段：出现任何新未知项 → 调用

planning

技能，重启流程链。