用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/sbusso/claudeclaw --skill install-extension命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
正在显示 SKILL.md
| name | install-extension |
| description | Install a ClaudeClaw extension (e.g., slack, triage) |
| trigger | /install <name> |
Install a ClaudeClaw extension from GitHub.
/install slack
/install triage
The argument is the extension short name (e.g., slack, triage). The full repo name is claudeclaw-<name>.
EXTENSION_NAME="${1}"
REPO="https://github.com/sbusso/claudeclaw-${EXTENSION_NAME}.git"
EXT_DIR="extensions/claudeclaw-${EXTENSION_NAME}"
[ -d "$EXT_DIR" ] && echo "Extension claudeclaw-${EXTENSION_NAME} is already installed." && exit 0
If already installed, ask if the user wants to update instead (git pull + rebuild).
mkdir -p extensions
git clone "$REPO" "$EXT_DIR"
cat "$EXT_DIR/manifest.json"
If manifest.json has a dependencies field, install them at the root:
cd "$(git rev-parse --show-toplevel)" && npm install <each dependency with version>
cd "$EXT_DIR" && npx tsc
If manifest.json has hooks.postInstall:
chmod +x "$EXT_DIR/hooks/install.sh"
bash "$EXT_DIR/hooks/install.sh" "$(git rev-parse --show-toplevel)"
This copies skills, agents, and agent skills into the ClaudeClaw root.
npm run build
Detect OS and restart:
macOS:
SERVICE_NAME=$(launchctl list | grep claudeclaw | awk '{print $3}')
[ -n "$SERVICE_NAME" ] && launchctl kickstart -k "gui/$(id -u)/$SERVICE_NAME"
Linux:
systemctl --user restart claudeclaw
Print confirmation:
Extension claudeclaw-<name> installed successfully.
- Skills: <list from manifest>
- Type: <channel|extension>
- Restart: service restarted
Add /compact command for manual context compaction. Solves context rot in long sessions by forwarding the SDK's built-in /compact slash command. Main-group or trusted sender only.
Add image vision to ClaudeClaw agents. Resizes and processes WhatsApp image attachments, then sends them to Claude as multimodal content blocks.
Add Ollama MCP server so the container agent can call local models for cheaper/faster tasks like summarization, translation, or general queries.
基于 SOC 职业分类