菜单
Diagnose and recover wedged shmerminal/openclaw sessions via the local shvix daemon. Use when the user reports a stuck terminal, frozen PTY, lockfile, port conflict, or corrupted session state.
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。
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。
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/thedotmack/Shmerminal --skill shvix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。
基于 SOC 职业分类
正在显示 SKILL.md
| name | shvix |
| description | Diagnose and recover wedged shmerminal/openclaw sessions via the local shvix daemon. Use when the user reports a stuck terminal, frozen PTY, lockfile, port conflict, or corrupted session state. |
| allowed-tools | Bash |
The user's symptom: $ARGUMENTS Active session id: ${CLAUDE_SESSION_ID}
shvix is the recovery agent. Your job is to forward the symptom to the local shvix daemon and report its response verbatim. Do not attempt to diagnose or fix the issue yourself — the daemon owns classification and runbook execution.
Run exactly this Bash block:
payload=$(jq -nc --arg s "$ARGUMENTS" --arg sid "$CLAUDE_SESSION_ID" \
'{symptom: $s, context: {session_id: $sid}}')
curl -sS --fail-with-body \
-X POST http://localhost:7749/fix \
-H 'content-type: application/json' \
-d "$payload"
Then:
message field is
the human-facing summary; action_taken is the runbook that ran;
requires_human: true means a human must take the next step.curl failure (connection refused, non-2xx, etc.): tell the user the
shvix daemon is not running and instruct them to start it with
shvix daemon in another terminal, then re-run /shvix <symptom>.
Do not retry. Do not try to fix the underlying issue without the daemon.classification is "unknown": report it and stop. Do not guess a
runbook. Human intervention is the correct outcome.