菜单
Flow Conductor for the aiops bundle. One entry — infers task type, shows plain-language steps, tracks journey in flow.state.yaml, dispatches skill + agent per phase.
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。
基于 SOC 职业分类
Delivery overlay — orchestrates lean ladder, tdd, prune, review with sub-phase gates; commits only on user approval.
Structured design process for the architect agent. Guides from grill conclusions to complete NOTES.md + tech-spec.md plus needed domain documentation updates through constraint gathering, module identification, interface design, alternative exploration, and risk analysis. Use when the architect needs to produce design decisions and technical specifications.
Build and query a persistent code graph for the target project. Uses graphify (Tree-sitter + Louvain clustering) for deterministic structural extraction, then the model adds semantic annotations. Use when the user says build graph, code graph, dependency graph, impact analysis, or when other skills need structured code understanding before proceeding.
Scan a codebase for deepening opportunities, present them as a visual HTML report, then grill through whichever one you pick. Use when the user says architecture review, codebase health, deepening, refactor suggestions, or complains about architectural friction.
Review in three modes — code (diff vs standards/spec), design (NOTES + tech-spec before planning), drift (implementation vs tech-spec before ship). Pick mode from Flow Conductor phase or user intent; do not mix artifact types.
Configure a target project for the aiops bundle — issue tracker, triage labels, domain doc layout. Silent defaults on inline bootstrap; full prompts on explicit /aiops-setup.
| name | aiops |
| description | Flow Conductor for the aiops bundle. One entry — infers task type, shows plain-language steps, tracks journey in flow.state.yaml, dispatches skill + agent per phase. |
| disable-model-invocation | true |
Single entry for the bundle. Ordinary users only need this command; experts can still /aiops <agent> ….
/aiops <what you want> # Conductor — infer, narrate, dispatch
/aiops 继续 / resume # Resume from .scratch/<slug>/flow.state.yaml
/aiops <agent-name> <task> # Expert — direct agent; still update journey if in a flow
Agents: architect, design-reviewer, planner, prototyper, builder, ui-designer, code-reviewer, quality-auditor, gitops
.scratch/<slug>/flow.state.yaml if user resumes or slug known. If current_phase_id is done, inform user the journey is complete. Else infer slug (kebab-case from description; user may override). See journey.md.docs/agents/ missing, run bootstrap inline:
aiops.yaml → silent defaults: local markdown issues + 1:1 triage labels (aiops-setup/aiops-yaml.md).aiops.yaml with issue_tracker.kind: github|gitlab → seed GitHub/GitLab tracker docs from yaml; do not prompt for tracker unless yaml incomplete.FlowState from journey + repo signals (see journey.md). Default delivery_mode: single_session unless user confirms multi or heuristics clearly warrant multi. Use flow_cli.py plan to generate the phase list.title_zh, body_zh, artifact_zh; English only in title_en. Hide skill/agent names unless user asks.agents/<agent>.md when agent set; invoke skill for the phase. Gather .scratch/<slug>/ inputs per agent Inputs.flow_cli.py validate --slug <slug> to verify gate artifacts. If validation fails, do not advance. On success, append gate name to gates_satisfied.flow_cli.py advance --slug <slug> to update current_phase_id and phases_done. On handoff, advance journey before writing temp handoff doc.delivery, hand off to /aiops-implement (owns lean → tdd → prune → review). Do not interleave those gates from the conductor.| Task type | Phases (abbrev) | Grill | Ends at |
|---|---|---|---|
| Feature | align → design → design_review → task_breakdown → delivery → drift_check → ship | yes | /gitops |
| Feature + UI | + ui_mockup before design_review | yes | /gitops |
| Bug | diagnose → delivery → drift_check → ship | no | /gitops |
| Incoming | triage → (align if unclear) → delivery → drift_check → ship | conditional | /gitops |
| Architecture health | graph_build → architecture_scan → align → design → design_review → task_breakdown → delivery → drift_check → ship | yes | /gitops |
| Prototype | prototype only | — | VERDICT.md |
| New personal skill | skill_authoring checklist | yes | new SKILL.md |
Dispatch rows for AGENTS.md live in skills/manifest.json → dispatch. Phase order detail: skills/aiops/scripts/phases.py.
Explore (opt-in): when explore_requested, insert explore phase after bootstrap (if any) and before task-specific phases.
Multi-session: after design_review → planning_prd → planning_issues → issue_session → delivery per issue. Single-session: after design_review → task_breakdown → delivery.
Prototype branch (optional): handoff → /prototype → handoff back; require VERDICT.md before planner/builder.
Lean is not active during grill/alignment phases.
Recommend multi-session when: 3+ modules, multiple slices, near smart zone, or AFK per-issue. Default single-session — single-session still includes task_breakdown (lightweight tasks.md); multi-session uses full planning_prd → planning_issues → issue_session.
SKILL.md — Cursor create-skill (External; docs/skill-registry.md)python3 <aiops-root>/skills/aiops/scripts/flow_cli.py {plan,init,advance,validate,satisfy-gate,phase-gates}CONTEXT.md; skill registry: docs/skill-registry.md