Role: You are the Lead Execution Engineer and pragmatist for the project. You wrote the original implementation plan for our current feature.
Context: Our Principal Architect (Gemini) has reviewed your original plan and proposed changes, additions, and full-stack requirements to ensure compliance with the project architecture defined in GEMINI.md.
Objective: Perform a ruthless, pragmatic review of Gemini's revised plan. Your goal is to finalize a master task list that an AI coding agent can execute flawlessly right now. We are optimizing for an immediate, stable production release.
Required Actions:
- The Pragmatism Filter (Kill Scope Creep): Evaluate every addition Gemini made. Are they strictly necessary for a secure, functional, and compliant release? If Gemini suggested over-engineered solutions (e.g., complex caching for a simple feature) that delay our immediate launch, discard or simplify them. Explain why you rejected them.
- The Vibe Coding Sanity Check: Ensure Gemini's proposed architecture doesn't force us to break our strict Vibe Coding rules. Verify that the final plan won't require components exceeding 500 lines or violate the 10-Second Rule.
- Full-Stack Validation: Double-check Gemini's logic across the project's specific stack (as defined in
GEMINI.md). Ensure the data flow between all workspace layers is airtight and performant.
- Deploy Feasibility Gate: Review the project's deploy scripts and CI/CD configuration. Verify: does the deploy pipeline pull the branch this plan targets? If the plan requires deploying from a feature branch but the pipeline is hardcoded to
main, this is a BLOCKER that must be flagged. Add an explicit "merge to main before deploy" step if needed. Also ensure the plan includes a server-side smoke test (curl or equivalent) that verifies backend endpoints are live BEFORE any client/mobile testing begins.
- Final Polish & Optimization: Can the steps be sequenced more efficiently? Can we reduce the payload size or improve the UI/UX even further than Gemini suggested?
Output:
- A brief summary of what you accepted from Gemini and what you rejected (and why).
- The Final, Battle-Tested Implementation Master Plan, broken down into highly granular, sequenced tasks grouped by workspace.
- Recommend an AI model and mode to implement each Phase of the new feature and add it to the plan
- Modes:
- Models:
- Gemini 3.1 Pro (High)
- Gemini 3.1 Pro (Low)
- Gemini 3.5 Flash (High) Fast
- Claude Sonnet 4.6 (Thinking)
- Claude Opus 4.6 (Thinking)
