Skip to content
Back to skills

Multi Phase Plan

ASecurity

Write a dependency plan with verifiable units and live checks.

  • 2 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added October 1, 2026
ai-agentsperformance

Security analysis

A100/100

Pro scans all 2 files and shows the line behind each finding

Scanned October 6, 2026

npx -y skills add williamwue/oh-my-stack --skill multi-phase-plan --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Multi Phase Plan?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Multi Phase Plan
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/williamwue-multi-phase-plan-oh-my-stack/badge)](https://www.skillsdirectory.com/skills/williamwue-multi-phase-plan-oh-my-stack)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: multi-phase-plan
description: "Write a dependency plan with verifiable units and live checks."
---

# Multi-phase plan

The plan is the deliverable. Use for a change with multiple dependent units or
pull requests; a small obvious edit does not need this ceremony. Inspect the
repository and resolve technical unknowns with bounded prototypes when
authorized. Ask only for choices that evidence cannot settle.

Write one section per independently verifiable unit: outcome, dependencies,
owned files or interfaces, build action, user-observable result, unit check,
matching-surface live check, and relevant performance rule. State the actual
baseline for a comparative metric. If the old surface lacks a feature, use an
absolute budget for added work instead of an invalid old/new ratio. A missing
control harness is a prerequisite or named risk, never a fictional test lane.

Choose the later execution owner explicitly: a bounded stack, separately
landed queue, or standing program. State who may integrate, merge, and publish.
For a standing execution program, name the verified scheduler, bounded wake
budget, and audit cadence, defaulting to one hour unless the task specifies
another interval. Describe durable pause behavior when that capability is
absent. The plan neither arms a wake nor starts execution. Reconcile progress
by side effects and report newly tracked changes rather than unchanged status.

Identify concurrent work only across non-overlapping ownership and capture
its base revisions. Use a checklist whose boxes require concrete evidence such
as a file, result, screenshot, or commit. Include prototypes, alternatives,
risks, and reading links. Validate paths and the dependency graph; remove
unresolved placeholders. [Technical-writing](../technical-writing/SKILL.md)
can tighten the prose.

Return the plan path or complete in-chat plan, unit order, dependencies,
verification rules, unresolved product choices, and explicit start condition.
Do not implement the plan merely because its design is finished.

Files in this skill

  • SKILL.md1.6 KB
  • agents/openai.yaml242 B

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…