CLI 命令

CLI 是 cc_orchestrator.py。

先设置 CC_ORCHESTRATOR_HOME：

export CC_ORCHESTRATOR_HOME="$HOME/.codex/skills/claude-code-orchestrator/scripts/cc-orchestrator"

Windows PowerShell：

$env:CC_ORCHESTRATOR_HOME = "$env:USERPROFILE\.codex\skills\claude-code-orchestrator\scripts\cc-orchestrator"

健康检查和发现

python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" selftest
python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" healthcheck
python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" list-profiles

启动 worker 前先跑这些。

工作区治理

初始化受管理的产物目录：

python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" init-workspace --cwd /path/to/project
python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" workspace-status --cwd /path/to/project
python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" folder-policy --cwd /path/to/project --apply

整理旧产物或噪声文件：

python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" migrate-data --cwd /path/to/project
python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" clean-workspace --cwd /path/to/project
python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" archive-runs --cwd /path/to/project --older-than-days 30
python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" repair-mcp-paths --cwd /path/to/project --create

这些命令只管理 Agent 生成的产物。会改路径或删除东西的动作默认先预览，除非传 --apply。

路由

只看路由，不启动 Claude Code：

python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" pick --role implementation --task-type complex_code

生成完整工作流计划：

python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" workflow-plan "Refactor this project safely"

模型评分和报告

python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" score-models
python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" write-auto-policy
python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" write-reports

write-auto-policy 会更新角色别名，让本机模型按角色自动选择。

write-reports 会把模型评分和策略报告写到 .agent-workspace/claude-code-orchestrator/reports。

启动 worker

只读架构 worker：

python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" run "Map this repository architecture" --role architecture

允许改文件的实现 worker：

python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" run "Fix the failing test in src/foo.py" --role implementation --allow-write --cwd /path/to/project

打开可视 Claude Code 窗口：

python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" run-visible "Inspect this project" --role architecture

查看运行结果

python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" last-run
python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" diff --cwd /path/to/project
python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" verify-run --run-id <run_id> --test-command "npm test"

run 会存到：

.agent-workspace/claude-code-orchestrator/runs/<run_id>/
  metadata.json
  prompt.txt
  stdout.txt
  stderr.txt

Windows 实时看 stdout：

Get-Content ".agent-workspace\claude-code-orchestrator\runs\<run_id>\stdout.txt" -Wait

macOS 或 Linux：

tail -f ".agent-workspace/claude-code-orchestrator/runs/<run_id>/stdout.txt"

实时控制

python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" run-streaming "Review this repo" --role review
python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" poll-run --run-id <run_id>
python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" summarize-run --run-id <run_id>
python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" compact-events --run-id <run_id>
python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" run-status
python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" stop-run --run-id <run_id> --force
python "$CC_ORCHESTRATOR_HOME/cc_orchestrator.py" mock-stream-test

poll-run 默认是 controller 模式，会返回压缩进度、风险、改动文件、时间线、工具调用摘要和 checkpoint 路径。

mock-stream-test 用假 Claude stream，不花模型额度也能测试 streaming。