plan

name	plan
description	Sequence work before execution. Use for planning requests like "break this down", "roadmap", "approach", "milestones", "how should we proceed", migration planning, rollout planning, and scoped next steps.
license	MIT
tags	["plan","roadmap","sequencing"]
metadata	{"author":"Oleg Shulyakov","version":"1.1.0","source":"github.com/olegshulyakov/agent.md","catalog":"utility","category":"project-management"}

plan

Turn a goal into a practical sequence of work.

---

Workflow

1. Define the goal, scope, and success condition.
2. Identify constraints, assumptions, dependencies, and unknowns.
3. Break work into ordered phases or steps.
4. Add risks and mitigation where failure would be costly.
5. Define verification for each meaningful phase.
6. End with the next concrete action.

---

Output

• Lead with the approach: state the overall strategy in one short paragraph.
• Use phases for larger work: include purpose, key tasks, dependencies, and validation.
• Keep steps scoped: each step should have a visible outcome.
• Flag blockers: name missing context that prevents a reliable plan.
• Avoid fake precision: do not invent owners, dates, or estimates without evidence.

---

Boundaries

Scenario: Work needs sequencing Given the user asks how to proceed Then identify phases, dependencies, assumptions, risks, verification, and immediate next actions

Scenario: Execution is already active Given the primary need is managing live owners, blockers, or handoffs Then do not treat planning as the primary behavior And route the output toward coordination only when the user asks to manage ongoing work

Scenario: Durable documentation is optional Given the user has not requested durable files Then default to a conversational plan And create durable files only when the work clearly needs durable task documentation

---

Error Paths

Scenario: Goal is unclear Given the goal is not clear enough to plan reliably Then ask the one question that most affects scope And provide a provisional outline if useful

Scenario: Many unknowns exist Given too many unknowns prevent reliable sequencing Then split discovery from execution And identify what must be learned first

Scenario: Execution is already active Given the user asks about ongoing work Then switch the output toward status, blockers, and handoff only when the user asks to manage that work

---

Verification

Scenario: Output passes quality check Given a plan has been produced Then later steps do not require missing earlier outputs And validation or acceptance checks are included for non-trivial work And the plan is revised when new facts change scope, risk, or sequence