
Retired. Local screen capture should use native macOS screencapture or shell directly.
ghost-os 桌面 computer-use 说明书。msgcode 直接暴露 `ghost_*` 原生工具;先读外部真相源 GHOST-MCP,再按本文的本地安装、探测和失败口径执行。
# memory skill 触发:用户要求“记住”“回忆”“检索历史记忆”。 优先入口:`~/.config/msgcode/skills/memory/main.sh` 默认 workspace:当前目录 `$PWD`(可被 `--workspace` 覆盖)。 常用: - `bash ~/.config/msgcode/skills/memory/main.sh add "用户偏好..." --json` - `bash ~/.config/msgcode/skills/memory/main.sh search "偏好" --limit 8 --json` - `bash ~/.config/msgcode/skills/memory/main.sh stats --json` - `bash ~/.config/msgcode/skills/memory/main.sh status --json`
This skill should be used when the model needs detailed image understanding beyond the system preview summary, and needs a provider-neutral capability index.
This skill should be used when the model needs to create, inspect, enable, disable, or remove recurring schedules in msgcode, or when diagnosing schedule state in the current workspace.
This skill should be used when a group or channel has multiple speakers and the model needs to identify the current speaker, maintain a workspace-local CSV table, or look up a sender by channel/chatId/senderId instead of guessing.
# file skill 触发:读取文件、查找内容、写文件、移动/复制文件。 优先入口:原生 `read_file` 与 `bash` 常用: - `read_file("/absolute/path/to/file")` - `rg --files . | head` - `sed -n '1,120p' README.md` - `cp ./AIDOCS/report.md ./AIDOCS/report.bak.md` 说明: - `file skill` 负责文件查找、读取、写入、复制、移动。 - 本地文件操作直接使用原生工具或 shell,不要再期待 `msgcode file ...` 包装层。 - 需要批量查找时优先 `rg` / `find`;需要局部读取时优先 `read_file` 或 `sed -n`。 - 如果要把文件回传到当前飞书会话,优先直接调用 `feishu_send_file(...)`,不要再走历史 `msgcode file send`。
# media skill(retired) 本 skill 已退役,不再进入 runtime 主索引。 本地截图请直接使用原生 macOS `screencapture` 或 shell,例如: - `mkdir -p ./AIDOCS/media/screenshots` - `screencapture -x -t png ./AIDOCS/media/screenshots/current.png` 不要再调用 `msgcode media screen`。
This skill should be used when the model needs real webpage access, long-form page text extraction or transcription, or to inspect and drive the Patchright browser CLI, verify Chrome root state, and diagnose browser instances and tabs in msgcode.
# thread skill 触发:查看/切换会话线程,读取线程消息。 优先入口:`msgcode thread ...` 常用: - `msgcode thread list --json` - `msgcode thread active --json` - `msgcode thread messages <thread-id> --limit 20 --json` - `msgcode thread switch <thread-id> --json`
# todo skill 触发:任务记录、查看待办、标记完成。 优先入口:`~/.config/msgcode/skills/todo/main.sh` 默认 workspace:当前目录 `$PWD`(可被 `--workspace` 覆盖)。 常用: - `bash ~/.config/msgcode/skills/todo/main.sh add "补充测试报告" --json` - `bash ~/.config/msgcode/skills/todo/main.sh list --json` - `bash ~/.config/msgcode/skills/todo/main.sh done <taskId> --json`
This skill should be used when the task involves an x.com or twitter.com post URL and the model needs to extract tweet text, images, videos, or thumbnails without guessing page content.
This skill should be used when the user explicitly wants to react to a message with an emoji and the current runtime already has a local reaction bridge endpoint available.
当用户明确要求主脑把一段编码、调研或审查任务委派给 codex / claude-code 这类子代理执行臂,并在完成后回收结果时使用。
This skill should be used when the task is to generate a short video with Google Veo through Gemini-compatible APIs, and the environment has a working GEMINI_API_KEY or GOOGLE_API_KEY.
用 Banana Pro API 生成、编辑、描述图片,并统一落盘到 AIDOCS 目录
需要把当前 workspace 里的文件回传到飞书会话时使用。优先直接调用 feishu_send_file,并让工具自己从 .msgcode/config.json 读取 runtime.current_chat_id;只有在必须显式确认 chatId 时,才读取配置文件,不要解析 session 文件名。
# gen skill 触发:图片生成、自拍、语音、音乐。 优先入口:`msgcode gen ...` 常用: - `msgcode gen image --prompt "赛博朋克猫" --json` - `msgcode gen selfie --ref ./AIDOCS/images/ref.png --json` - `msgcode gen tts --text "你好" --json` - `msgcode gen music --prompt "lofi" --json`
This skill should be used when the task is complex enough to benefit from task-local planning files, but should not introduce a new plan mode, supervisor, or memory layer.
This skill should be used when the task needs structured web extraction with Scrapling, including simple fetch, dynamic rendering, or stealthy fetching, and ordinary page reading is not enough.
This skill should be used when the model needs detailed image understanding through a local LM Studio vision model and needs the actual local calling contract.