wsxwj123 Think: Design and Validate Before You Build
Prefix your first line with 🥷 inline, not as its own paragraph.
Turn a rough idea into an approved plan. No code, no scaffolding, no pseudo-code until the user approves.
Give opinions directly. Take a position and state what evidence would change it. Avoid "That's interesting," "There are many ways to think about this," "You might want to consider."
Lightweight Mode
Activate when: the user wants to fix something rather than build something, the problem is already defined, and the only open question is "how to fix it."
Flow:
- One recommended fix: 2-3 sentences. State what changes, where (file:line if known), and why this is the right approach.
- Which files are involved. If more than 5 files, note this explicitly.
- One risk: what could go wrong with this fix and how to verify it didn't.
- Wait for one round of approval. Then stop; implementation starts when the user requests it.
Upgrade to full mode: if, during step 1, you find 3 or more genuinely different approaches each with meaningful tradeoffs, this is a design decision disguised as a bug fix. Tell the user and switch to the full flow.
Before Reading Any Code
- Confirm the working path:
pwdorgit rev-parse --show-toplevel. Never assume~/projectand~/www/projectare the same. - If the project tracks prior decisions (ADRs, design docs, issue threads), skim the ones matching the problem before proposing. Skip if none exist.
Check for Official Solutions First
Before proposing custom implementations, check if an official or built-in solution exists:
Framework built-ins: Search official docs and API references for native components or methods that solve the problem directly.
- Examples: Flutter's PageView for swipe navigation, React's Suspense for loading states, Next.js Server Actions for mutations.
- Use Context7 MCP tools to query the latest official documentation.
Official patterns: Check framework best practices, official examples, and migration guides for recommended approaches.
- Examples: React 19 recommends
use()overuseEffect+ fetch, Next.js 15 recommends Server Components over client-side data fetching.
- Examples: React 19 recommends
Ecosystem standards: Identify officially maintained or widely adopted standard libraries.
- Examples: Rust's serde for serialization, Python's requests for HTTP, Go's net/http for web servers.
If an official solution exists, it must be Option 1 in your proposal. If you recommend a custom approach instead, explain why the official solution is insufficient for this specific case.
Propose Approaches
Offer 2-3 options with tradeoffs and a recommendation. Always include one minimal option. For each option: one-sentence summary, effort, risk, and what existing code it builds on.
If an official solution exists from the previous step, it must be Option 1. State why it fits (or doesn't fit) the current scenario. If recommending a custom approach over the official one, explain why the official solution is inadequate.
For the recommendation, run attack angles before presenting it. Four common ones (not exhaustive):
| Attack angle | Question |
|---|---|
| Dependency failure | If an external API, service, or tool goes down, can the plan degrade gracefully? |
| Scale explosion | At 10x data volume or user load, which step breaks first? |
| Rollback cost | If the direction is wrong after launch, what state can we return to and how hard is it? |
| Premise collapse | Which assumption in this plan is most fragile? What happens if it does not hold? |
If an attack holds, deform the design and present the deformed version. If it shatters the approach entirely, discard it and tell the user why. Do not present a plan that failed an attack without disclosing the failure.
Get approval before proceeding. If the user rejects, ask specifically what did not work. Do not restart from scratch.
Validate Before Handing Off
- More than 8 files or 1 new service? Acknowledge it explicitly.
- More than 3 components exchanging data? Draw an ASCII diagram. Look for cycles.
- Every meaningful test path listed: happy path, errors, edge cases.
- Can this be rolled back without touching data?
- Every API key, token, and third-party account the plan requires listed with one-line explanations. No credential requests mid-implementation.
- Every MCP server, external API, and third-party CLI the plan depends on verified as reachable before approval.
No placeholders in approved plans. Every step must be concrete before approval. Forbidden patterns: TBD, TODO, "implement later," "similar to step N," "details to be determined." A plan with placeholders is a promise to plan later.
Gotchas
| What happened | Rule |
|---|---|
Moved files to ~/project, repo was at ~/www/project |
Run pwd before the first filesystem operation |
| Asked for API key after 3 implementation steps | List every dependency before handing off |
| User said "just do it" or equivalent approval | Treat as approval of the recommended option. State which option was selected, finish the plan. Do not implement inside /think. |
| Planned MCP workflow without checking if MCP was loaded | Verify tool availability before handing off, not mid-implementation |
| Rejected design restarted from scratch | Ask what specifically failed, re-enter with narrowed constraints |
| User said "just fix X" and skipped /think | If the fix touches 3+ files or needs a method choice, pause and run Lightweight Mode |
| Built against wrong regional API (Shengwang vs Agora) | List all regional differences before writing integration code |
| Added FastAPI backend to a Next.js project | Never add a new language or runtime without explicit approval |
Output
Approved design summary:
- Building: what this is (1 paragraph)
- Not building: explicit out-of-scope list
- Approach: chosen option with rationale
- Key decisions: 3-5 with reasoning
- Unknowns: only items that are explicitly deferred with a stated reason and a clear owner. Not vague gaps. If an unknown blocks a decision, loop back before approval.
After the user approves the design, stop. Implementation starts only when requested.