plan

Decompose an agreed design into ordered, independently shippable delivery steps.

Updated Feb 12, 2026
One-click install
npx skills add https://github.com/kieranpotts/skills --skill plan-kieranpotts
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: plan
Source: https://github.com/kieranpotts/skills/tree/main/skills/plan
Command: npx skills add https://github.com/kieranpotts/skills --skill plan-kieranpotts

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Turning an agreed design into working code often fails because the work is tackled in the wrong order or in chunks too large to review, test, and revert safely. This Skill breaks a design into a sequenced checklist of small, independently mergeable steps so delivery supports continuous integration. ## Core Features & Use Cases - Thinnest-first slicing: Identifies the smallest end-to-end slice (walking skeleton, riskiest integration, or flagged path) to anchor the plan. - Risk-ordered decomposition: Schedules steps carrying unknowns before merely laborious work, with each step tagged by mode (AFK or HITL), pass/fail signal, and any flag, fixture, or migration dependency. - Plan store integration: Discovers the project's design, specification, and plan stores and writes the plan in the store's own format without touching any source code. - Use Case: After agreeing on a design for a new /orders endpoint with idempotency, invoke this Skill to produce a numbered checklist from route scaffolding through migration, handler, idempotency, flag-gated rollout, and flag removal. ## Quick Start Ask the agent to break the agreed design into a sequenced implementation plan of small, independently mergeable steps.

Frequently Asked Questions about plan

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I break a software design into implementation steps?▼

Start from the agreed design and its acceptance criteria, find the thinnest end-to-end slice first, then split the remaining work into steps that each merge, test, and revert independently. Order steps by risk so unknowns are resolved before laborious work begins.

How to sequence development tasks for continuous integration?▼

Schedule the riskiest integrations and unproven assumptions first, keep easy polish and copy work last, and give every step an observable pass/fail signal such as a test name or command. Each step should be reviewable in under 30 minutes and never more than a working day.

When should I not use a delivery planning skill?▼

Skip it when the work is genuinely a single atomic step, when no design has been agreed yet, or when you need architectural decisions made. The plan skill only decomposes an agreed design; it does not write requirements, make design choices, or implement code.

Does the plan skill modify source code or configuration?▼

No. The skill is explicitly planning-only and must not change any source file, configuration, schema, or dependency manifest. Its only output is a numbered checklist written into the project's own plan store.

How are feature flags handled in a delivery plan?▼

Use a feature flag when it lets a step ship independently without exposing incomplete work, and name the flag on the step where it is used. The plan must include a final step that removes the flag once the feature has fully landed.