rabbitcli-shared
Core conventions for calling the rabbitcli CLI from an AI Agent — output contract, safety rules, environments. Auto-loaded alongside any other rabbitcli skill.
来源信息
- 仓库
- RabbitAI-Lab/RabbitAITest
- 最近来源活动
- 2026年9月30日 02:13
- 检测到的 SKILL.md 语言
- 英语
- 星标
- 0
- 分支
- 0
安装方式
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
检查来源文件
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
正在显示 SKILL.md
SKILL.md
来源说明 · 只读预览- name
- rabbitcli-shared
- description
- Core conventions for calling the rabbitcli CLI from an AI Agent — output contract, safety rules, environments. Auto-loaded alongside any other rabbitcli skill.
# rabbitcli shared rules
## Output contract
- Success: stdout, exit 0, `{"ok": true, "data": ...}`.
- Failure: stderr, exit != 0 (usage errors exit 2), `{"ok": false, "error": {"type": ..., "message": ..., "hint": ...}}`.
- Judge success by `ok == true` or the exit code — never by `code == 0`.
## Before calling
1. Check config: `rabbitcli config show` (note `_dir` and `_env`).
2. Discover usage: `rabbitcli schema` and `rabbitcli schema <group>`.
3. For any command with side effects, run with `--dry-run` first and show the preview to the user.
## Safety rules
- Never pass `--format pretty` when consuming output programmatically; use default JSON.
- Never print tokens. The CLI masks them; keep it that way.
- Use `--env` to point at a dev/staging backend when testing; confirm with the user before running against prod.
## Common flags
`--format`, `--env`, `--config-dir`, `--dry-run`, `--page-all`, `--verbose` / `--debug` (logs go to files, never stdout).
在 GitHub 查看