Skip to content
Back to skills

Plan Task

ASecurity

Turn a request into a short, structured task plan before starting work - Context, Higher Goal, Acceptance Criteria, Out of Scope, Summary of Changes - and use the same structure for the PR description at the end. Use when starting any non-trivial task, when the user asks to "plan", "scope" or "break down" work, or when opening a pull request.

  • 12 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added October 15, 2026
ai-agentsgodatabase

Security analysis

A100/100

Scanned October 15, 2026

npx -y skills add ymatrix1/vantage --skill plan-task --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Plan Task?

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

Security grade badge for Plan Task
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/ymatrix1-plan-task/badge)](https://www.skillsdirectory.com/skills/ymatrix1-plan-task)

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: plan-task
description: Turn a request into a short, structured task plan before starting work - Context, Higher Goal, Acceptance Criteria, Out of Scope, Summary of Changes - and use the same structure for the PR description at the end. Use when starting any non-trivial task, when the user asks to "plan", "scope" or "break down" work, or when opening a pull request.
---

# Plan a task

A short plan up front prevents the two classic agent failures: solving the wrong problem, and quietly growing the scope. The same five sections become the PR description, so the plan is checked against the result.

## The template

```markdown
## Context
The problem, why it matters now, and the current state. "Blocked by: none" or what must land first.

## Higher Goal
What this serves beyond the task itself, in one or two lines.

## Acceptance Criteria
- [ ] A verifiable outcome, with how to check it (a test, a command, a measurement)
- [ ] ...

## Out of Scope
What this task deliberately doesn't do, and where it goes instead.

## Summary of Changes
TBD (filled in when done: files changed, tests added, how it was proven)
```

## Steps

1. **Ask before assuming** if the goal, the expected behavior or a limit is unclear. One short question beats a wrong plan.
2. **Write the plan** with the template. Keep it under about 20 lines.
   - Each acceptance criterion is an **outcome you can check**, not a step ("search returns results in under 200 ms", not "add an index").
   - Out of Scope names at least one tempting extra you won't do.
3. **Show the plan** and get a quick OK for anything bigger than a small fix.
4. **Put it where the project tracks work:** the project's task tracker if it has one, otherwise the PR description or the top of the handoff note.
5. **While working:** tick criteria as they're proven (see `prove-it`). Anything new you discover goes into a follow-up, not into this task.
6. **At the end:** fill in Summary of Changes and use the whole template as the PR description, with every criterion ticked or explained.

## Example

> ## Context
> Login checks hit the database on every request; p95 latency is 180 ms. Blocked by: none.
>
> ## Higher Goal
> Make the app feel instant on every page.
>
> ## Acceptance Criteria
> - [ ] p95 of the session check under 50 ms in the load test (`npm run bench:auth`)
> - [ ] Revoked sessions stop working within 1 minute (new test)
>
> ## Out of Scope
> Changing the login page; caching other endpoints.
>
> ## Summary of Changes
> TBD

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…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.