
Think: Design Before You Build
Turns rough ideas into approved, decision-complete plans before any code is written.
- Repository
- tw93/Waza
- Stars
- 7,493
- Licence
- MIT
- Files
- 4 files
What it does
When installed, the agent stops jumping straight into code when you describe a feature or problem. Instead it asks for approval on a plan first: the goal, constraints, chosen approach, rejected alternatives, tests, and handoff steps. It checks existing project docs, config values and official solutions before recommending a design, and it gives a clear opinion rather than a list of vague options. It can also switch to an evaluation mode for keep-or-remove decisions and a triage mode for judging a batch of requests.
Great for
- Developers planning a new feature or architecture before writing code
- Teams deciding whether an idea is worth building
- People who want a written handoff plan an agent or teammate can execute without re-deciding
Skip it if
- Simple bug fixes or small edits, which the skill itself says it is not for, since the extra planning step adds delay
- Non-technical users who only want finished output such as a document or design, since the skill is about planning, not producing deliverables
Try asking
- “Plan this: add offline sync to our notes app, what approach should we take?”
- “Should we keep the legacy export feature, or is it worth removing?”
- “How should I design a rate limiter for our public API?”
- “Here are five feature requests from users, which are worth doing?”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 4 original files, the MIT licence and an attribution note.
think.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip think.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip think.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- Give the agent the real repo context and point it at your AGENTS.md or CLAUDE.md so it can check project rules before planning.
- Answer the approval question explicitly; the skill is designed to wait for your sign-off before any implementation.
- Ask for a lightweight plan when the problem is already defined, and a full plan when you have several competing approaches.
What's inside
- references/durable-context.md
- references/mode-evaluation.md
- references/mode-triage.md
- SKILL.md
Redistributed unmodified from tw93/Waza under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.