aios-workflow-router
AIOS Workflow Router
This skill routes your task to the appropriate workflow. Invoke this BEFORE any other action.
Quick Decision Tree
用户请求 → 任务类型判断 → 调用相应技能/流程
Task Type Detection
1. Design/Creative Tasks (设计/创意任务)
Keywords: 设计、创意、新功能、新特性、build、create、implement、设计、创意、brainstorm、头脑风暴
Route to: Planning workflow
- Use
superpowers:brainstormingif available, otherwise follow the brainstorming process below - Create design doc in
docs/plans/ordocs/superpowers/specs/ - Get user approval before implementation
2. Debug/Failure Tasks (调试/故障任务)
Keywords: 调试、bug、错误、失败、error、fail、debug、修复、fix、不工作、broken
Route to: Debugging workflow
- Use
superpowers:systematic-debuggingif available - Gather evidence first (logs, error messages, screenshots)
- Form hypothesis, test, iterate
3. Multi-step/Long-running Tasks (多步骤/长任务)
Keywords: 长任务、多步骤、harness、checkpoint、evidence、long-running、multi-step、复杂任务、orchestrat
Route to: Harness workflow
- Use
aios-long-running-harnessskill - Set preflight budgets and stop conditions
- Create checkpoints with evidence
4. Parallel/Agent Team Tasks (并行/团队任务)
Keywords: 并发、并行、agent team、团队、多agent、多个独立、dispatch、parallel、concurrent
Route to: Parallel dispatch workflow
- Identify independent problem domains
- Confirm there are at least 2 truly independent domains before dispatching
- Use
superpowers:dispatching-parallel-agentsif available - If no subagent tool available, emulate with explicit task queues
- While waiting, emit heartbeat progress every ~30s; if no worker completes after ~120s, stop waiting and fall back to sequential execution
5. Implementation Tasks (实现任务)
Keywords: 实现、implement、开发、develop、编码、code、写代码
Route to: Implementation workflow
- Check if plan exists in
docs/plans/ - If no plan, go through brainstorming first
- Use
superpowers:test-driven-developmentif available - Implement with evidence checkpoints
6. Analysis Tasks (分析任务)
Keywords: 分析、analysis、研究、research、investigate、调查、为什么、why
Route to: Analysis workflow
- Gather information from codebase, logs, history
- Document findings
- Present recommendations
- Default to single-agent execution; do not dispatch explorer/parallel agents unless the user explicitly asks for delegation or parallel work
Workflow Execution
Standard Flow
1. Route → 2. Plan → 3. Execute → 4. Verify → 5. Complete
Mandatory Steps
- Route: Identify task type (this skill)
- Plan: Create or reference plan document
- Execute: Follow plan with checkpoints
- Verify: Assert completion with evidence
- Complete: Update docs and commit
Fallback Behaviors
If superpowers:* skills are not available (plugin not installed):
- brainstorming: Use the built-in brainstorming process in this skill
- systematic-debugging: Use the debugging process below
- dispatching-parallel-agents: Execute sequentially or emulate with task queues
- verification-before-completion: Manually verify artifacts exist
Built-in Brainstorming Process
When superpowers:brainstorming is not available:
- Explore context - Check files, docs, recent commits
- Ask clarifying questions - One at a time, understand purpose
- Propose approaches - 2-3 options with trade-offs
- Present design - Get user approval
- Write plan - Save to
docs/plans/YYYY-MM-DD-<topic>.md
Built-in Debugging Process
When superpowers:systematic-debugging is not available:
- Gather evidence - Logs, errors, screenshots
- Form hypothesis - Based on evidence
- Test hypothesis - Make minimal change
- Verify fix - Confirm with evidence
- Document - Update runbook if needed
Completion Gate
Before claiming any task complete:
- Target action succeeded
- Expected artifacts exist
- Evidence documented
- No regressions introduced
Resource Links
memory/specs/行为规范.json- Safety specificationsmemory/specs/风险检测.json- Risk detection rulesdocs/plans/- Implementation plansmemory/history/- Operation records
More from rexleimo/rex-cli
skill-creator
Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, update or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.
3contextdb-autopilot
Use when running tasks in Codex CLI, Claude Code, Gemini CLI, or opencode and you need automatic context persistence (init/session/event/checkpoint/context-pack) plus interactive auto-routing without manual contextdb commands.
3seed2-manga-drama
当用户要把单图或创意脚本做成AI漫剧短视频时使用。提供Seed2.0风格的四阶段流程:分镜脚本、主角设定、分镜生成、视频验收,并输出可直接投喂生成模型的结构化提示词。
3debug
Evidence-first runtime debugging for application bugs, regressions, flaky behavior, and unclear failures. Use when an agent is asked to debug an issue and should avoid speculative fixes by forming hypotheses, attaching to or starting a logging session, instrumenting code, collecting runtime logs, analyzing the recorded log file, applying only proven fixes, and verifying the result before removing instrumentation, especially for browser or frontend issues where logs should go directly to the active collector endpoint instead of app-local proxy APIs.
2find-skills
Helps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express interest in extending capabilities. This skill should be used when the user is looking for functionality that might exist as an installable skill.
2xhs-ops-methods
当用户要学习或执行小红书运营方法时使用。提供可复用的“定方向-定人设-写排发互-复盘增长”流程,支持多账号协作,并强调人工审核与合规发布。
2