team-setup

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

team-setup

Overview

概述

team-setup

is an interactive skill that guides you through setting up the team AI directives. It presents four modes, explains each option, confirms your choice, and executes the setup.

It is invoked in two ways:

User-invoked (
```
/team-setup
```
) — anytime, to configure or check a project.
Model-invoked by
team-boot
— automatically at session start when a project has no
```
.adlc/init-options.json
```
configuration (self-install), so an unconfigured project wires itself without the user knowing the command.

The skill is non-destructive: it never overwrites existing files or directories. If the target path already contains a configured team AI directives, it detects this and offers the "Already configured" mode instead.

team-setup

是一个交互式 skill，可指导你完成团队 AI 指令的设置。它会提供四种模式，解释每个选项，确认你的选择并执行设置。

它有两种调用方式：

用户调用（
```
/team-setup
```
）——可随时调用，用于配置或检查项目。
由
team-boot
模型调用——当项目在会话开始时没有
```
.adlc/init-options.json
```
配置（自安装），便会自动执行。因此，未配置的项目无需用户了解该命令即可完成接入。

该 skill 不会造成破坏性操作：绝不会覆盖已有文件或目录。如果目标路径中已配置团队 AI 指令，它会检测到这一点，并改为提供“已配置”模式。

When to Use

使用时机

Starting a new team from scratch and need a neutral team AI directives scaffold to fill in later.
Your team already has a directives repo on GitHub and you want to clone it locally.
You have a local team AI directives directory already (e.g., from a previous project) and want to wire it up.
You're unsure whether the team AI directives is already configured and want a quick check.
When the project isn't yet wired to a team AI directives (no
```
.adlc/init-options.json
```
```
team_ai_directives
```
field).
Automatically via
```
team-boot
```
when it detects an unconfigured project at session start (self-install).

从零开始组建新团队，需要一个中立的团队 AI 指令脚手架，以便稍后填写。
团队在 GitHub 上已有指令仓库，希望将其克隆到本地。
已经在本地拥有团队 AI 指令目录（例如来自之前的项目），希望将其接入项目。
不确定团队 AI 指令是否已配置，希望快速检查。
项目尚未接入团队 AI 指令（
```
.adlc/init-options.json
```
中没有
```
team_ai_directives
```
字段）。
```
team-boot
```
在会话开始时检测到项目未配置，自动触发此流程（自安装）。

Decline Handling (when model-invoked by team-boot)

拒绝处理（由 team-boot 模型调用时）

When

team-boot

invokes this skill because the project is unconfigured, the user may choose not to set up team AI directives right now. Handle decline explicitly to avoid a re-prompt loop:

If the user declines at mode selection, do not run any mode. Exit cleanly and tell
```
team-boot
```
the user declined.
Offer a persistent opt-out: "Don't ask again for this project?" On yes (build mode only), write
```
.adlc/init-options.json
```
with
```
team_ai_directives: null
```
:
bash
```
echo '{"team_ai_directives": null}' > ".adlc/init-options.json"
```
This marker makes
```
team-boot
```
skip setup silently on every future prompt.
In plan/read-only mode, a persistent opt-out cannot be written — the decline is session-scoped only; tell
```
team-boot
```
to defer.
Never force a mode; the setup is user-consented at every step.

当

team-boot

因项目未配置而调用此 skill 时，用户可以选择暂时不设置团队 AI 指令。应明确处理拒绝操作，以避免重复提示循环：

如果用户在模式选择阶段拒绝，请勿运行任何模式。干净退出，并告知
```
team-boot
```
用户已拒绝。
提供永久退出选项：“不再为此项目询问？”如果用户选择“是”（仅限 build 模式），则写入
```
.adlc/init-options.json
```
，并设置
```
team_ai_directives: null
```
：
bash
```
echo '{"team_ai_directives": null}' > ".adlc/init-options.json"
```
此标记会让
```
team-boot
```
在今后的每次提示中静默跳过设置。
在 plan/只读模式下，无法写入永久退出标记；拒绝仅在当前会话范围内生效；告知
```
team-boot
```
暂缓处理。
绝不能强制选择某种模式；设置的每一步都必须获得用户同意。

Core Process

核心流程

Goal

目标

Set up a team AI directives using one of four modes.

使用四种模式之一设置团队 AI 指令。

Security: Input Validation (all modes)

安全：输入校验（所有模式）

Before executing any mode, validate every user-supplied value (paths, URLs, team names). These values are interpolated into shell commands; unvalidated input is a command-injection vector.

Paths (
```
{DEST}
```
,
```
{ABSOLUTE_PATH}
```
): reject if they contain any of
```
`
```
,
```
$
```
,
```
;
```
,
```
|
```
,
```
&
```
,
```
(
```
,
```
)
```
,
```
<
```
,
```
>
```
, newline, or backslash. Resolve to an absolute path with
```
realpath
```
/
```
Resolve-Path
```
before use.
Team name: must match
```
^[A-Za-z0-9 ._-]+$
```
. Reject anything else.
Clone URL (Mode 1): must start with
```
https://
```
. Reject
```
file://
```
,
```
ssh://
```
, and any non-
```
https
```
scheme unless the user explicitly confirms the risk. Cloning runs no code from the repo, but the cloned content is read by agents later — only clone repositories you trust.

If any value fails validation, report which value and why, and re-ask. Never interpolate a user value into a Python/eval source string — pass it through the environment (see Mode 2).

Fast path: if

.adlc/init-options.json

already contains a valid

team_ai_directives

path, skip to Mode 4 (Already Configured) — do not re-clone, re-point, or re-scaffold.

执行任何模式之前，校验每个用户提供的值（路径、URL、团队名称）。这些值会被插入 shell 命令；未经校验的输入会成为命令注入入口。

路径（
```
{DEST}
```
、
```
{ABSOLUTE_PATH}
```
）：如果包含以下任一字符，则拒绝处理：
```
`
```
、
```
$
```
、
```
;
```
、
```
|
```
、
```
&
```
、
```
(
```
、
```
)
```
、
```
<
```
、
```
>
```
、换行符或反斜杠。使用前通过
```
realpath
```
/
```
Resolve-Path
```
将其解析为绝对路径。
团队名称：必须匹配
```
^[A-Za-z0-9 ._-]+$
```
。拒绝其他所有内容。
克隆 URL（模式 1）：必须以
```
https://
```
开头。拒绝
```
file://
```
、
```
ssh://
```
以及任何非
```
https
```
协议，除非用户明确确认风险。克隆操作本身不会运行仓库中的代码，但 Agent 之后会读取克隆内容，因此只应克隆受信任的仓库。

如果任何值未通过校验，请说明是哪个值以及失败原因，然后重新询问。绝不能将用户提供的值插入 Python/eval 源代码字符串——应通过环境变量传递（参见模式 2）。

**快速路径：**如果

.adlc/init-options.json

已包含有效的

team_ai_directives

路径，则直接跳至模式 4（已配置）——不要重新克隆、重新指向或重新创建脚手架。

Modes at a Glance

模式概览

#	Mode	When	Details in
1	Clone from GitHub	team already has a directives repo	`references/mode-clone.md`
2	Point to existing local path	directives dir already exists locally	`references/mode-local.md`
3	Scaffold new empty directives	starting a team from scratch	`references/mode-scaffold.md`
4	Already configured	check / verify existing wiring	`references/mode-configured.md`

#	模式	使用场景	详情见
1	从 GitHub 克隆	团队已有指令仓库	`references/mode-clone.md`
2	指向现有本地路径	本地已存在指令目录	`references/mode-local.md`
3	创建新的空团队 AI 指令	从零开始组建团队	`references/mode-scaffold.md`
4	已配置	检查或验证现有接入	`references/mode-configured.md`

Mode 1: Clone from GitHub

模式 1：从 GitHub 克隆

Full walkthrough in

references/mode-clone.md

完整操作步骤见

references/mode-clone.md

。

Mode 2: Point to Existing Local Path

模式 2：指向现有本地路径

Full walkthrough in

references/mode-local.md

完整操作步骤见

references/mode-local.md

。

Mode 3: Scaffold New Empty team AI directives

模式 3：创建新的空团队 AI 指令

Full walkthrough in

references/mode-scaffold.md

完整操作步骤见

references/mode-scaffold.md

。

Mode 4: Already Configured

模式 4：已配置

Full walkthrough in

references/mode-configured.md

完整操作步骤见

references/mode-configured.md

。

Mode Selection Flow

模式选择流程

Explore: Present the user with four options:

How would you like to set up team-ai-directives?

1) Clone from GitHub — Clone an existing repository
2) Point to existing local path — Use a team AI directives you already have
3) Scaffold new empty team AI directives — Create a fresh neutral team AI directives
4) Already configured — Check existing configuration

Present: For the chosen mode, explain what will happen and show details.
Confirm: Ask the user to confirm before executing.
Write/Execute: Perform the setup for the chosen mode.

探索：向用户提供四个选项：

How would you like to set up team-ai-directives?

1) Clone from GitHub — Clone an existing repository
2) Point to existing local path — Use a team AI directives you already have
3) Scaffold new empty team AI directives — Create a fresh neutral team AI directives
4) Already configured — Check existing configuration

展示：针对所选模式说明将要执行的操作并展示具体细节。
确认：执行前要求用户确认。
写入/执行：按所选模式执行设置。

Script Entry Points (deterministic, headless use)

脚本入口（确定性、headless 使用）

The Mode 1/2 validation and wiring logic has ONE home — the shared helper scripts — and is exposed as flags so headless callers (e.g.

adlc-cli team setup --directives <path>

) run the same checks without an LLM session:

scripts/bash/team-helpers.sh [--validate DIR] [--wire DIR]

scripts/powershell/team-helpers.ps1 [-Validate DIR] [-Wire DIR]

--validate DIR

runs the Verification structure check (exit 0 when complete, exit 1 with

MISSING:

lines when not).

--wire DIR

validates, then merges the absolute path into the project's

.adlc/init-options.json

team_ai_directives

(existing keys preserved; refuses to wire an invalid structure). Both twins behave identically. Interactive sessions still go through Mode Selection above; the in-session path and the headless path share the same implementation.

模式 1/2 的验证和接入逻辑只有一个实现位置——共享辅助脚本——并以 flags 的形式公开，使 headless 调用方（例如

adlc-cli team setup --directives <path>

）无需 LLM 会话也能运行相同检查：

scripts/bash/team-helpers.sh [--validate DIR] [--wire DIR]

scripts/powershell/team-helpers.ps1 [-Validate DIR] [-Wire DIR]

--validate DIR

会运行 Verification 结构检查（完成时退出码为 0；未完成时退出码为 1，并输出

MISSING:

行）。

--wire DIR

会先进行验证，然后将绝对路径合并到项目的

.adlc/init-options.json

中，作为

team_ai_directives

（保留已有键；拒绝接入无效结构）。两个对应脚本的行为完全一致。交互式会话仍通过上述模式选择流程进行；会话内路径与 headless 路径共享同一实现。

Post-Setup Configuration

设置后配置

After any mode completes successfully, update the project configuration:

Write

team_ai_directives

.adlc/init-options.json

Verify the team AI directives is accessible by running a quick health check:
- ```
{TEAM_AI_DIRECTIVES}/context_modules/constitution.md
```
  exists
- ```
{TEAM_AI_DIRECTIVES}/.skills.json
```
  exists and is valid JSON
Check the
```
.gitignore
```
convention (layout allowlist — see the Gitignore Convention Check step below).
Inject the project-level
```
AGENTS.md
```
directive — full command and managed-section contract in
```
references/post-setup-agents.md
```
.
Install MCP config — copy-once contract (seed the project's
```
.mcp.json
```
only when absent; an existing project file wins and is never merged into) in
```
references/post-setup-mcp.md
```
.

任何模式成功完成后，更新项目配置：

将

team_ai_directives

写入

.adlc/init-options.json

通过快速健康检查验证团队 AI 指令可正常访问：
- ```
{TEAM_AI_DIRECTIVES}/context_modules/constitution.md
```
  存在
- ```
{TEAM_AI_DIRECTIVES}/.skills.json
```
  存在且是有效 JSON
检查
```
.gitignore
```
约定（布局允许列表——参见下方 Gitignore 约定检查步骤）。
注入项目级
```
AGENTS.md
```
指令——完整命令与托管区块约定见
```
references/post-setup-agents.md
```
。
安装 MCP 配置——仅复制一次的约定（仅在项目不存在
```
.mcp.json
```
时进行种子写入；已有项目文件优先，且绝不合并写入），详见
```
references/post-setup-mcp.md
```
。

Gitignore Convention Check

Gitignore 约定检查

Verify

.gitignore

follows the ADLC allowlist — a fresh

team setup

must never leave a wholesale

.adlc/

ignore in place (it would hide the tracked

.adlc/

artifacts the team model relies on):

.gitignore

contains a bare

.adlc/

rule, replace it (and any

.adlc/*

!.adlc/...

lines that conflict) with the R7 allowlist: ignore

.adlc/*

.adlc/evals/results/

.adlc/memory/*

.adlc/team-learn-report.md

.adlc/team-levelup-report.md

, and

graphify-out/

, while re-including

!.adlc/init-options.json

!.adlc/workspace.yml

!.adlc/drafts/

!.adlc/evals/

!.adlc/memory/

!.adlc/memory/evals/

, and

!.adlc/memory/evals/holdout.json

. The canonical rule list lives in the workspace skill's

scripts/bash/paths.sh

(

GITIGNORE_RULES_ALLOWLIST

, mirrored in

scripts/powershell/paths.ps1

Also verify the agent-install surface rules are ignored (
```
.agents/
```
,
```
.opencode/
```
,
```
.claude/
```
,
```
.cursor/
```
,
```
.codex/
```
,
```
.gemini/
```
,
```
.qwen/
```
,
```
.devin/
```
,
```
.tabnine/
```
,
```
skills-lock.json
```
,
```
.skills.json
```
,
```
.events.json
```
,
```
.pytest_cache/
```
,
```
.ruff_cache/
```
). Note that
```
.mcp.json
```
is deliberately not in this list: it is committed project source — human-edited, never generated — so a
```
.gitignore
```
entry for it is a defect to report, not a missing rule to add; its guardrail is variable names only, never literal secrets —
```
env
```
indirection is the guard, not
```
.gitignore
```
.
Never remove unrelated (non-ADLC) rules; only add missing allowlist rules.
Tracked legacy files keep working — the allowlist gates only untracked files, so no migration is required before the switch.

验证

.gitignore

是否遵循 ADLC 允许列表——全新执行

team setup

后，绝不能保留整体忽略

.adlc/

的规则（否则会隐藏团队模型依赖的已跟踪

.adlc/

产物）：

如果

.gitignore

包含纯粹的

.adlc/

规则，则将其（以及任何与之冲突的

.adlc/*

!.adlc/...

行）替换为 R7 允许列表：忽略

.adlc/*

、

.adlc/evals/results/

、

.adlc/memory/*

、

.adlc/team-learn-report.md

、

.adlc/team-levelup-report.md

和

graphify-out/

，同时重新包含

!.adlc/init-options.json

、

!.adlc/workspace.yml

、

!.adlc/drafts/

、

!.adlc/evals/

、

!.adlc/memory/

、

!.adlc/memory/evals/

和

!.adlc/memory/evals/holdout.json

。规范规则列表位于 workspace skill 的

scripts/bash/paths.sh

（

GITIGNORE_RULES_ALLOWLIST

，并在

scripts/powershell/paths.ps1

中同步维护）。

同时验证 Agent 安装目录相关规则已被忽略（
```
.agents/
```
、
```
.opencode/
```
、
```
.claude/
```
、
```
.cursor/
```
、
```
.codex/
```
、
```
.gemini/
```
、
```
.qwen/
```
、
```
.devin/
```
、
```
.tabnine/
```
、
```
skills-lock.json
```
、
```
.skills.json
```
、
```
.events.json
```
、
```
.pytest_cache/
```
、
```
.ruff_cache/
```
）。请注意，
```
.mcp.json
```
被有意排除在此列表之外：它是需要提交的项目源文件，由人工编辑且绝不自动生成，因此为它添加
```
.gitignore
```
条目是需要报告的缺陷，而不是需要补充的缺失规则；其防护机制仅涉及变量名称，绝不涉及明文 secret——真正的防护是
```
env
```
间接引用，而不是
```
.gitignore
```
。
绝不能删除与 ADLC 无关的规则；只添加缺失的允许列表规则。
已跟踪的旧文件仍可继续使用——允许列表仅限制未跟踪文件，因此切换前无需执行迁移。

Common Rationalizations

常见错误认知

Rationalization	Why it's wrong	What to do instead
"I'll just clone it manually."	Manual cloning skips the `.adlc/init-options.json` wiring, so agents won't find the team AI directives.	Use Mode 1 — it clones AND configures.
"I already have a team AI directives directory, I'll just use it."	The directory may be incomplete (missing required files) or not wired in config.	Use Mode 2 — it validates the structure and creates the config entry.
"I'll just create a few files by hand."	An incomplete scaffold breaks health checks and agent discovery.	Use Mode 3 — it creates all 10 required files with valid structure.
"I'm sure it's already configured."	The path may be stale, moved, or the env var may point to a deleted dir.	Use Mode 4 — it validates the existing configuration.
"Scaffolding without a team name is fine."	The team name is used in `README.md` — a blank name makes the team AI directives anonymous and harder to audit.	Always provide a team name in Mode 3.

错误认知	为什么是错误的	应该怎么做
“我手动克隆一下就行。”	手动克隆会跳过 `.adlc/init-options.json` 的接入配置，因此 Agent 无法找到团队 AI 指令。	使用模式 1——它会克隆并完成配置。
“我已经有团队 AI 指令目录，直接使用即可。”	该目录可能不完整（缺少必需文件），或者尚未接入配置。	使用模式 2——它会验证结构并创建配置条目。
“我手动创建几个文件就行。”	不完整的脚手架会破坏健康检查和 Agent 发现机制。	使用模式 3——它会创建全部 10 个必需文件，并确保结构有效。
“我确定已经配置好了。”	路径可能已失效、被移动，或者环境变量可能指向一个已删除的目录。	使用模式 4——它会验证现有配置。
“创建脚手架时不提供团队名称也可以。”	团队名称会用于 `README.md` ；名称为空会使团队 AI 指令失去身份标识，并且更难审计。	在模式 3 中始终提供团队名称。

Red Flags

危险信号

Cloning over an existing directory — Mode 1 refuses if the destination already exists to prevent overwrites.
Pointing to a non-existent path — Mode 2 validates the path exists before proceeding.
Scaffolding without required dirs being writable — Mode 3 creates directories with
```
mkdir -p
```
but will fail on permission errors; check permissions first.
Skipping the
team_ai_directives
config write — without this field in
```
init-options.json
```
, agents cannot discover the team AI directives.
Using a relative path in
init-options.json
— always resolve to an absolute path so the config is portable across working directories.
Skipping
git init
in Mode 3 — a scaffolded team AI directives without git cannot be used by
```
/team-levelup
```
(branch/commit/PR flow). Mode 3 runs
```
git init
```
automatically; if you skip it, run
```
git init
```
manually before
```
/team-levelup
```
.
Skipping the project-level AGENTS.md injection — without the
```

```
managed section in the project's
```
AGENTS.md
```
, agents without event support have no session-start instruction to load team context. The
```
.adlc/init-options.json
```
config alone is insufficient — it tells skills where the team AI directives is, but nothing tells the agent to check. (For agents with event support, the session-start hook injects the orientation regardless, but AGENTS.md remains the fallback and the source of the Team Context in Use output contract.)
Interpolating user input into Python/shell source strings — pass paths through the environment (
```
os.environ
```
) instead; string interpolation of
```
$ABSOLUTE_PATH
```
into a Python one-liner is a command-injection vector.
Cloning a non-
https://
URL in Mode 1 — reject
```
file://
```
/
```
ssh://
```
/other schemes; cloned content is read by agents later, so only clone trusted repos.
Skipping the MCP config install — a project with no
```
.mcp.json
```
of its own is never seeded and loses access to team-declared MCP servers; a project that has one only loses the gap report (its file is never written either way).
Accepting shell metacharacters in paths or team names — validate before interpolating into
```
mkdir
```
/
```
git commit
```
/heredocs (see Input Validation).
Treating user decline as an error — declining setup is a valid outcome; exit cleanly, tell
```
team-boot
```
the user declined, and offer the
```
team_ai_directives: null
```
opt-out marker (build mode only).
Writing the opt-out marker in plan/read-only mode — a persistent opt-out requires a write; in plan mode the decline is session-scoped and setup defers instead.

克隆到已有目录——模式 1 会在目标已存在时拒绝执行，以防止覆盖。
指向不存在的路径——模式 2 会在继续之前验证路径是否存在。
创建脚手架时必需目录不可写——模式 3 使用
```
mkdir -p
```
创建目录，但遇到权限错误时会失败；请先检查权限。
跳过写入
team_ai_directives
配置——如果
```
init-options.json
```
中没有此字段，Agent 将无法发现团队 AI 指令。
在
init-options.json
中使用相对路径——始终将路径解析为绝对路径，以确保配置可跨工作目录使用。
在模式 3 中跳过
git init
——没有 git 的脚手架式团队 AI 指令无法用于
```
/team-levelup
```
（分支/提交/PR 流程）。模式 3 会自动运行
```
git init
```
；如果跳过，则必须在
```
/team-levelup
```
之前手动运行
```
git init
```
。
跳过注入项目级
AGENTS.md
——如果项目的
```
AGENTS.md
```
中没有
```

```
托管区块，不支持 event 的 Agent 就无法获得会话启动时加载团队上下文的指令。仅有
```
.adlc/init-options.json
```
配置并不足够——它会告诉 skill 团队 AI 指令位于何处，却没有任何信息指示 Agent 去检查它。（对于支持 event 的 Agent，会话启动 hook 无论如何都会注入初始方向说明，但
```
AGENTS.md
```
仍是回退方案，也是 Team Context in Use 输出约定的来源。）
将用户输入插入 Python/shell 源代码字符串——应通过环境变量传递路径（
```
os.environ
```
）；将
```
$ABSOLUTE_PATH
```
插入 Python 单行代码字符串会形成命令注入入口。
在模式 1 中克隆非
https://
URL——拒绝
```
file://
```
/
```
ssh://
```
/其他协议；克隆内容之后会被 Agent 读取，因此只应克隆受信任的仓库。
跳过安装 MCP 配置——如果项目自身没有
```
.mcp.json
```
，则不会为其进行种子写入，并且会失去对团队声明的 MCP 服务器的访问权限；如果项目已有该文件，则只会失去差异报告（无论哪种情况都不会写入其文件）。
接受路径或团队名称中的 shell 元字符——在插入
```
mkdir
```
/
```
git commit
```
/heredoc 之前进行校验（参见输入校验）。
将用户拒绝视为错误——拒绝设置是有效结果；应干净退出，告知
```
team-boot
```
用户已拒绝，并提供
```
team_ai_directives: null
```
退出标记选项（仅限 build 模式）。
在 plan/只读模式下写入退出标记——永久退出需要写权限；在 plan 模式下，拒绝仅在当前会话范围内生效，设置操作应暂缓执行。

Verification

验证

The team AI directives directory exists at the configured path.

{TEAM_AI_DIRECTIVES}/context_modules/constitution.md

exists.

{TEAM_AI_DIRECTIVES}/context_modules/rules/

exists.

{TEAM_AI_DIRECTIVES}/context_modules/personas/

exists.

{TEAM_AI_DIRECTIVES}/context_modules/examples/

exists.

```
{TEAM_AI_DIRECTIVES}/CDR.md
```
exists.
```
{TEAM_AI_DIRECTIVES}/.skills.json
```
exists and is valid JSON.

.adlc/init-options.json

contains a

team_ai_directives

field with the absolute path.

.gitignore

follows the layout allowlist (no wholesale

.adlc/

ignore; tracked exceptions

!.adlc/init-options.json

!.adlc/workspace.yml

!.adlc/drafts/

!.adlc/evals/

!.adlc/memory/evals/holdout.json

present).

Project-level
```
AGENTS.md
```
exists and contains the
```

```
managed section with the event-hook awareness note, fallback
```
team-boot
```
invocation, and the Team Context in Use output contract.

(Mode 3 only)

git rev-parse --is-inside-work-tree

succeeds inside

{TEAM_AI_DIRECTIVES}

Running
```
team-verify
```
(Phase 0 of team-repair) passes all 7 checks.
All user-supplied paths/URLs/team names passed Input Validation (no shell metacharacters; clone URL is
```
https://
```
).
Mode 2 wrote
```
team_ai_directives
```
via the environment (no
```
$ABSOLUTE_PATH
```
interpolation into Python source).
If
```
{TEAM_AI_DIRECTIVES}/.mcp.json
```
exists: the project's
```
.mcp.json
```
was seeded from it only when the project had none — an existing project file was left untouched (never merged into or rewritten), any gap between the two was reported rather than closed, and unresolved env vars were highlighted.
(Model-invoked by
```
team-boot
```
) a user decline exited cleanly without running any mode; the persistent opt-out was offered, and
```
team_ai_directives: null
```
was written only in build mode.

{TEAM_AI_DIRECTIVES}/context_modules/constitution.md

存在。

{TEAM_AI_DIRECTIVES}/context_modules/rules/

存在。

{TEAM_AI_DIRECTIVES}/context_modules/personas/

存在。

{TEAM_AI_DIRECTIVES}/context_modules/examples/

存在。

```
{TEAM_AI_DIRECTIVES}/CDR.md
```
存在。
```
{TEAM_AI_DIRECTIVES}/.skills.json
```
存在且是有效 JSON。

.adlc/init-options.json

包含

team_ai_directives

字段，且值为绝对路径。

.gitignore

遵循布局允许列表（不存在整体忽略

.adlc/

的规则；已跟踪例外

!.adlc/init-options.json

、

!.adlc/workspace.yml

、

!.adlc/drafts/

、

!.adlc/evals/

、

!.adlc/memory/evals/holdout.json

均存在）。

项目级
```
AGENTS.md
```
存在，并包含
```

```
托管区块，其中含有 event hook 感知说明、回退
```
team-boot
```
调用和 Team Context in Use 输出约定。

（仅限模式 3）在

{TEAM_AI_DIRECTIVES}

中运行

git rev-parse --is-inside-work-tree

成功。

运行
```
team-verify
```
（team-repair 的阶段 0）时，全部 7 项检查均通过。
所有用户提供的路径、URL 和团队名称均通过输入校验（不包含 shell 元字符；克隆 URL 使用
```
https://
```
）。
模式 2 通过环境变量写入
```
team_ai_directives
```
（未将
```
$ABSOLUTE_PATH
```
插入 Python 源代码）。
如果
```
{TEAM_AI_DIRECTIVES}/.mcp.json
```
存在：仅当项目原本没有
```
.mcp.json
```
时，才从该文件进行种子写入；已有项目文件保持不变（绝不合并写入或重写），两者之间的差异仅被报告而不被消除，并且未解析的环境变量会被重点提示。
（由
```
team-boot
```
模型调用时）用户拒绝后已干净退出且未运行任何模式；已提供永久退出选项，并且仅在 build 模式下写入
```
team_ai_directives: null
```
。

Configuration

配置

```
TEAM_AI_DIRECTIVES
```
— Path to the team AI directives (overrides
```
.adlc/init-options.json
```
).

.adlc/init-options.json

— Project-level config file with

team_ai_directives

field.

Default fallback:
```
team-ai-directives/
```
relative to project root.
```
team-helpers.sh
```
/
```
team-helpers.ps1
```
— Shared scripts used for scaffolding and path resolution.

```
TEAM_AI_DIRECTIVES
```
——团队 AI 指令的路径（优先于
```
.adlc/init-options.json
```
）。
```
.adlc/init-options.json
```
——包含
```
team_ai_directives
```
字段的项目级配置文件。
默认回退路径：相对于项目根目录的
```
team-ai-directives/
```
。
```
team-helpers.sh
```
/
```
team-helpers.ps1
```
——用于创建脚手架和解析路径的共享脚本。

12-Factor Alignment

12-Factor 对齐

Factor XI (Directives as Code) — establishes a version-controlled team directives repository.

因子 XI（Directives as Code）——建立一个纳入版本控制的团队指令仓库。