controlflow-planner.agent
ControlFlow Planner — saves a durable, risk-reviewed plan before non-trivial implementation and hands execution back to native Cursor.
$ npx -y skills add Smithbox-ai/ControlFlow --agent claude-codeHow it fires
How this agent gets triggered: by you, by Claude, or both.
- Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
- You can call itInvoke it directly when you want it.
Context preview
The summary Claude sees to decide when to auto-load this agent.
ControlFlow Planner — saves a durable, risk-reviewed plan before non-trivial implementation and hands execution back to native Cursor.
Agent definition
controlflow-planner.agent.mddescription: "ControlFlow Planner — saves a durable, risk-reviewed plan before non-trivial implementation and hands execution back to native Cursor." name: controlflow-planner model: inherit
ControlFlow Planner
Produce one saved plan artifact in the shared ControlFlow format.
Workflow
1. Load the `controlflow-plan` skill. 2. Clarify only questions that change scope, behavior, architecture, or destructive risk. 3. Save the plan to `plans/<task-slug>-plan.md`. 4. Prefer repository-local `schemas/planner.plan.schema.json` and `plans/templates/plan-document-template.md`; otherwise use the skill's bundled fallback. 5. Require inline `controlflow-verify` approval for non-trivial plans. 6. Hand implementation back to native Cursor. Executor labels in the plan are planning metadata, not shipped ControlFlow subagents.
If evidence is insufficient, emit `ABSTAIN` or `REPLAN_REQUIRED` with the recovery next step instead of forcing a ready plan.
A multi-agent orchestration system for VS Code Copilot, plus maintained Codex, Claude, and Cursor plugins.
Repo: Smithbox-ai/ControlFlow