菜单
Compatibility command for the streamlined unattended developer workflow. Plans a software task with sl-plan when needed, then routes execution through sl-run in mode:unattended. Prefer sl-plan followed by sl-run for new workflows.
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/akornmeier/super-looper --skill lfg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。
Run an explicit, non-mutating compatibility diagnostic for the super-looper plugin. Use only when validating a Claude Code or Codex installation, runtime adapter, bundled reference and script resolution, interactive question support, and one-worker subagent dispatch.
Compatibility command for interactive implementation. Plans a software task when needed, then routes canonical Markdown execution through sl-run in mode:interactive. Prefer sl-run directly for new workflows.
Compatibility command for the streamlined unattended developer workflow. Plans a software task with sl-plan when needed, then routes execution through sl-run in mode:unattended. Prefer sl-plan followed by sl-run for new workflows.
基于 SOC 职业分类
正在显示 SKILL.md
| name | lfg |
| description | Compatibility command for the streamlined unattended developer workflow. Plans a software task with sl-plan when needed, then routes execution through sl-run in mode:unattended. Prefer sl-plan followed by sl-run for new workflows. |
| argument-hint | [feature description | plan:<path> | state:<path>] |
lfg is a compatibility wrapper. Tell the user once, concisely: "lfg now routes through sl-plan -> sl-run; use those commands directly for new workflows." Then perform the route instead of recreating the former pipeline in this skill.
STRATEGY.md or the active plan. A goal change requires interactive sl-strategy or a human-approved plan revision before a new run. The supervisor and goal guard report violations as exit 8, typed_failure: "goal-drift".profile:, run-id:, base-ref:, state-path:, and max-workers: tokens when forwarding to sl-run.sl-run owns workflow transitions and authority gates.sl-run stops at durable review_ready by default. It never infers engineer approval or delivery authority.mode:legacy-pipeline is present, remove that token, read references/legacy-pipeline.md, and follow it. This explicit compatibility route exists only for the legacy per-phase loop driver; never select it implicitly.state:<absolute-path> is present, invoke sl-run with the state marker, mode:unattended, and the preserved forwarding tokens. Do not plan or initialize another run.plan:<repo-relative-path> is present, invoke sl-run with that exact plan marker, mode:unattended, and the preserved forwarding tokens. Let sl-run validate the canonical Markdown plan; do not silently re-plan an invalid path.sl-plan with the software-task description in pipeline/unattended context. It must produce canonical Markdown. If planning rejects a non-software task or does not return a readable plan path, stop with that error.sl-run with plan:<returned-path> mode:unattended plus the preserved forwarding tokens.sl-run's compact status and absolute state path. Emit <promise>DONE</promise> as the final non-empty line only when sl-run durably reports review_ready or completed; otherwise stop honestly without the sentinel.progress:<path> and resume:<path> belong only to mode:legacy-pipeline. Never translate them into new primary-workflow state. New unattended runs use state-path:<absolute-path> and resume with state:<absolute-path>.