Compile Agent 上下

文档社区

作者仓库星标 0
作者仓库 skills-registry

解读按原文结构重写，命令、链接、术语均保留；右侧可核对作者原始 SKILL.md

方法与流程

Workflow：原文以此为独立章节。

适用与边界

当前原文没有单列适用场景。
5. Configure auto-maintenance rule (optional)：If autowatch=true, create/update .cursor/rules/agents-maintainer.md. Define a rule that prompts AGENTS refresh when significant project structure changes occur.

原文中的明确线索

要点：「Use this skill to run the old init command workflow as a Codex-native skill.」、「- Read root project files (README, manifests, config files).」、「- Search for agents in .cursor/agents, .cline/agents, .agents, agents, .ai, prompts, system-prompts.」、「- Capture coding and formatting standards from lint/format configs.」、「- Write a complete AGENTS.md in repository root.」、「- If autowatch=true, create/update .cursor/rules/agents-maintainer.md.」
文件与命令：init、scandepth、shallow、medium、deep、includedependencies、true、false

流狐整理：以上内容来自当前 SKILL.md 的章节与原词；未补写作者没有声明的工具、兼容性或能力。

Compile Agents Context

Overview

Use this skill to run the old init command workflow as a Codex-native skill. It standardizes AGENTS discovery and regeneration with explicit options.

Inputs

scan_depth: shallow, medium, or deep (default medium)
include_dependencies: true or false (default true)
auto_watch: true or false (default true)

Workflow

1. Discover project context

Read root project files (README, manifests, config files).
Scan directories according to scan_depth:
- shallow: root + first-level key folders.
- medium: default, includes feature/service folders.
- deep: broad scan for full architecture and conventions.

2. Discover agents and skills

Search for agents in .cursor/agents, .cline/agents, .agents, agents, .ai, prompts, system-prompts.
Search for skills in .cursor/skills, .codex/skills, .agents/skills, and user-level paths if accessible.
Capture each discovered item with name, location, purpose, and trigger scenarios.

3. Extract conventions

Capture coding and formatting standards from lint/format configs.
Capture architecture, testing, and build/deploy patterns from project files and docs.
Include dependency context only when include_dependencies=true.

4. Generate or refresh AGENTS.md

Write a complete AGENTS.md in repository root.
Include project overview, stack, structure, agent ecosystem, skill registry, conventions, and workflows.
Preserve existing high-signal custom instructions when regenerating.

5. Configure auto-maintenance rule (optional)

If auto_watch=true, create/update .cursor/rules/agents-maintainer.md.
Define a rule that prompts AGENTS refresh when significant project structure changes occur.

Reference

Use /Users/carloscaceres/local/marius/code_samples/credit-applications-hub/.agents/skills/context-compilation-guide/SKILL.md for the detailed section template and generation depth guidance.

Output Contract

Updated /Users/carloscaceres/local/marius/code_samples/credit-applications-hub/AGENTS.md
Optional /Users/carloscaceres/local/marius/code_samples/credit-applications-hub/.cursor/rules/agents-maintainer.md
Short summary listing files updated and discovery statistics

Source: cmcp0/credit-applications-hub — distributed by TomeVault.

流狐档案作者与许可取自来源；运行、权限和网络为流狐检测或估算

流狐分类: 文档
作者声明 Agent: 未找到明确声明；不据此推断已兼容或已测试
静态检查: 88 / 100 · 启发式扫描，不代表运行安全
作者 / 版本 / 许可: @tomevault-io · 未声明 license
流狐 Token 估算: 低消耗
流狐接入估算: 即装即用
是否需要外部 API Key: 未发现要求
检测到的系统要求: 未声明
底层运行要求: 未声明
检测到的文件与系统行为: 只读
允许写入 / 修改
检测到的网络行为: 仅限本地
安装命令数: 无（仅作为资料）

档案由构建时根据 SKILL.md 与安装命令自动衍生，可能与作者实际意图存在差异。

需要注意： 未限定 allowed-tools，默认拥有全部工具权限。

输出预览 compile-agents-context.preview

作者没有在当前 SKILL.md 中定义固定输出样例。

SKILL.md SKILL.md

# Compile Agents Context

## Overview

Use this skill to run the old `init` command workflow as a Codex-native skill. It standardizes AGENTS discovery and regeneration with explicit options.

## Inputs

- `scan_depth`: `shallow`, `medium`, or `deep` (default `medium`)
- `include_dependencies`: `true` or `false` (default `true`)
- `auto_watch`: `true` or `false` (default `true`)

## Workflow

### 1. Discover project context

- Read root project files (`README`, manifests, config files).
- Scan directories according to `scan_depth`:
  - `shallow`: root + first-level key folders.
  - `medium`: default, includes feature/service folders.
  - `deep`: broad scan for full architecture and conventions.

### 2. Discover agents and skills

- Search for agents in `.cursor/agents`, `.cline/agents`, `.agents`, `agents`, `.ai`, `prompts`, `system-prompts`.
- Search for skills in `.cursor/skills`, `.codex/skills`, `.agents/skills`, and user-level paths if accessible.
- Capture each discovered item with name, location, purpose, and trigger scenarios.

### 3. Extract conventions

- Capture coding and formatting standards from lint/format configs.
- Capture architecture, testing, and build/deploy patterns from project files and docs.
- Include dependency context only when `include_dependencies=true`.

### 4. Generate or refresh AGENTS.md

- Write a complete `AGENTS.md` in repository root.
- Include project overview, stack, structure, agent ecosystem, skill registry, conventions, and workflows.
- Preserve existing high-signal custom instructions when regenerating.

### 5. Configure auto-maintenance rule (optional)

- If `auto_watch=true`, create/update `.cursor/rules/agents-maintainer.md`.
- Define a rule that prompts AGENTS refresh when significant project structure changes occur.

…

证据边界与执行链路

作者原文负责流程事实；流狐只索引当前章节、要点、文件与命令。

章节    -> Overview → Inputs → Workflow → 1. Discover project context → 2. Discover agents and skills → 3. Extract conventions
要点    -> Use this skill to run the old init command workflow as a Codex-native skill. · - Read root project files (README, manifests, config files). · - Search for agents in .cursor/agents, .cline/agents, .agents, agents, .ai, prompts, system-prompts. · - Capture coding and formatting standards from lint/format configs. · - Write a complete AGENTS.md in repository root. · - If autowatch=true, create/update .cursor/rules/agents-maintainer.md.
文件/命令 -> init · scandepth · shallow · medium · deep · includedependencies · true · false
内容 SHA-256 -> b7ae971c444f