- Home
- Skills
- Obra
- Superpowers
- Writing Plans
writing-plans_skill
- Shell
36.9k
GitHub Stars
1
Bundled Files
3 weeks ago
Catalog Refreshed
2 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstart where the catalogue uses aiagentskills.
npx veilstart add skill obra/superpowers --skill writing-plans- SKILL.md3.2 KB
Overview
This skill generates detailed, test-first implementation plans from a spec or requirements document before any coding begins. It produces step-by-step, bite-sized tasks that assume the engineer has no repository context and limited test design skill. Plans include exact file paths, full code snippets, commands with expected output, and a clear save location so work can be executed or handed off reliably.
How this skill works
I inspect the spec or feature description and decompose it into atomic actions (2–5 minute steps) that follow TDD: write a failing test, run it, implement minimal code, run tests, commit. The plan always includes a required markdown header, exact file create/modify instructions, complete code examples, test commands with expected results, and explicit commit commands. At the end I save the plan to docs/plans/YYYY-MM-DD-<feature-name>.md and present execution handoff options.
When to use it
- Before writing any code for a multi-step feature or refactor
- When the implementer has little to no context about the codebase
- When you need reproducible, reviewable work units for parallel development
- When enforcing TDD and frequent, small commits is required
- When handing work to remote contributors or junior engineers
Best practices
- Start the plan with: "I'm using the writing-plans skill to create the implementation plan."
- Keep each step a single, verifiable action (2–5 minutes) following TDD
- Use exact file paths and complete code listings, not vague instructions
- Prefer minimal implementations to satisfy tests; avoid speculative features (YAGNI)
- Make frequent commits with clear messages and include expected test output in the plan
Example use cases
- Add a new API endpoint with request validation and tests, with exact files and test cases
- Implement a core algorithm change by first defining failing unit tests and minimal fixes
- Split a large feature into parallel tasks for multiple contributors, each with its own commitable steps
- Onboard a new engineer by giving them a complete, executable plan for their first contribution
- Convert an informal spec into a repeatable worktree saved plan for automated execution
FAQ
Save the completed plan to docs/plans/YYYY-MM-DD-<feature-name>.md using the exact filename pattern.
How granular should steps be?
Each step should be one concrete action taking 2–5 minutes: write one failing test, run it, implement minimal code, run tests, commit.
How do I hand off execution?
Offer two options: Subagent-Driven (stay in session, fresh subagent per task) or Parallel Session (new session running executing-plans in a separate worktree). Reference required sub-skills in the header.