orchestrate
Pipeline orchestration: dispatch the highest-priority ready tasks/work units to agents, manage capacity, and coordinate the Todo to Done flow. Invok…
它会碰到什么
这一栏是扫描器报的事实,不是结论。命中多不等于有毒(安全工具、规则库、示例脚本本来就会包含危险写法),命中少也不等于干净。它和你手上的凭据、文件、网络有什么关系,需要你自己看。
技能内容
> Invoked as /agiflow:orchestrate. In hosts without slash-prompts, this skill is triggered by matching intent and drives AgiFlow via its MCP tools.
Usage:
/agiflow:orchestrate- Check pipeline state and dispatch the next highest-priority task
Guardrails
- This is a read-assess-dispatch loop, not an implementation prompt.
- Do NOT implement tasks here — use
/agiflow:run-taskfor that. - Keep capacity checks honest: do not dispatch if at capacity.
- Report pipeline state even when no dispatch is needed.
AgiFlow Project Management Guidelines
Follow the shared AgiFlow project-management guidelines in [references/agiflow-agents.md](../../references/agiflow-agents.md) — agent assignment, the task status workflow and transitions, work-unit best practices, and the tags strategy apply to this workflow.
Steps
Track these steps as TODOs and complete them one by one.
1. Assess Current Pipeline State
- Use
list_tasksto count tasks in each active status column:
status: "In Progress"— tasks currently being coded by agentsstatus: "Testing"— tasks running test suitesstatus: "Review"— tasks awaiting human reviewstatus: "Blocked"— tasks requiring human interventionstatus: "Todo"— tasks ready for pickup (sorted by priority automatically)
- Report the pipeline state in a concise table:
Pipeline State:
┌──────────────┬───────┐
│ Status │ Count │
├──────────────┼───────┤
│ Todo │ N │
│ In Progress │ N │
│ Testing │ N │
│ Review │ N │
│ Blocked │ N │
└──────────────┴───────┘
2. Check Capacity
- Determine active task count:
In Progress+Testingcombined. - Check capacity limit (default: 3 concurrent active tasks unless specified).
- If at or above capacity:
- Report: "At capacity (N active tasks). No dispatch needed."
- List any Blocked tasks that need human attention.
- Stop here.
3. Prioritize the Todo Queue
- Use
list_taskswithstatus: "Todo"to retrieve the ready queue.
- Tasks are automatically sorted by priority (high → medium → low).
- Within the same priority, tasks are ordered by position then creation date.
- Review the top candidates:
- Show the top 3-5 Todo tasks with: slug, title, priority, assignee, retryCount (from devInfo)
- Skip any task where
devInfo.retryCount >= devInfo.maxRetries(should be Blocked — flag it)
4. Dispatch
- Pick the highest-priority eligible Todo task.
- Report the dispatch decision:
Dispatching: [SLUG] Task title (priority: high)
Reason: Highest priority task in Todo queue
- Instruct the agent to run the task:
- Use
/agiflow:run-task <slug>to execute it - Or if already in an agent session, invoke the run-task prompt directly
5. Surface Blocked Tasks
- If there are any Blocked tasks, list them with their
blockedReasonfrom devInfo:
Blocked Tasks Requiring Human Attention:
- [SLUG] Task title: <blockedReason>
- Suggest actions for each blocked task (e.g., resolve dependency, provide credentials, clarify spec).
Common Mistakes to Avoid
- ❌ Dispatching when already at capacity
- ❌ Picking a lower-priority task when a higher-priority Todo exists
- ❌ Dispatching a task that has
retryCount >= maxRetries(it should be Blocked) - ❌ Skipping the pipeline state report
- ❌ Attempting to implement the task inside this prompt
想直接用这个技能?
本站把开放许可(MIT / Apache 等)的技能按仓库打包整理到网盘,点一下转存到你自己的网盘,不用一个个从 GitHub 拉。许可未声明的技能只给原始仓库链接,不打包。
它属于哪个仓库
plugins/AgiFlow/ai-plugin/skills/orchestrate/SKILL.md同一个仓库里的其他技能
同名技能的其他版本
有 2 个不同仓库或目录里都有叫 orchestrate 的技能。它们内容并不相同,别混用:
- hashgraph-online/awesome-codex-plugins — Trigger: active multi-agent run. Handles inbox reading, dep resolution, message formatting