菜单
Execute quality-gated multi-step workflows with evaluator checks, approval gates, and typed results. Orchestrates Claude Code and Codex CLI skills into reproducible flows with auto-retry on failure.
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。
| name | c8c-workflow-runner |
| description | Execute quality-gated multi-step workflows with evaluator checks, approval gates, and typed results. Orchestrates Claude Code and Codex CLI skills into reproducible flows with auto-retry on failure. |
| version | 1.0.0 |
| metadata | {"openclaw":{"requires":{"bins":["node"]},"emoji":"⚡","homepage":"https://github.com/anthropics/c8c"}} |
Use this skill when you want OpenClaw to execute a deterministic workflow instead of improvising the entire task in a single agent session.
Use the packaged CLI compatibility path when:
Do not use it when:
npm install -g @c8c-ai/clilobster --help (the lobster binary is registered globally after install)lobster run --mode tool /abs/path/workflow.yaml --args-json '{"input":"...","inputType":"text","projectPath":"/abs/path/project","provider":"claude"}'
If the result returns needs_approval, resume with:
lobster resume --token '<resume-token>' --approve yes
lobster resume --token '<resume-token>' --approve no
If the checkpoint should be resolved outside OpenClaw, use:
c8c-workflow hil list --project /abs/path/project
c8c-workflow hil show --task '<task-token>'
c8c-workflow hil approve --task '<task-token>'
ok: false: treat as a runtime or contract errorstatus: "cancelled": the approval was explicitly rejectedstatus: "needs_approval": the workflow is safe to resume later; do not restart it from scratchPost-ship documentation update. Reads all project docs, cross-references the diff, updates README/ARCHITECTURE/CONTRIBUTING/CLAUDE.md to match what shipped, polishes CHANGELOG voice, cleans up TODOS, and optionally bumps VERSION.
CEO/founder-mode plan review. Rethink the problem, find the 10-star product, challenge premises, expand scope when it creates a better product. Three modes: SCOPE EXPANSION (dream big), HOLD SCOPE (maximum rigor), SCOPE REDUCTION (strip to essentials).
Designer's eye review of a live site. Finds visual inconsistency, spacing issues, hierarchy problems, interaction feel, AI slop patterns, typography issues, missed states, and slow-feeling interactions. Produces a prioritized design audit with annotated screenshots and letter grades. Infers your design system and offers to export as DESIGN.md. Report-only — never modifies code. For the fix loop, use /qa-design-review instead.
Eng manager-mode plan review. Lock in the execution plan — architecture, data flow, diagrams, edge cases, test coverage, performance. Walks through issues interactively with opinionated recommendations.
Report-only QA testing. Systematically tests a web application and produces a structured report with health score, screenshots, and repro steps — but never fixes anything. Use when asked to "just report bugs", "qa report only", or "test but don't fix". For the full test-fix-verify loop, use /qa instead.
Pre-landing PR review. Analyzes diff against the base branch for SQL safety, LLM trust boundary violations, conditional side effects, and other structural issues.