β Drop in a sentence. Get back a researched, dependency-ordered, audit-clean product plan β with ready-to-paste designer and engineering prompts. β
π If this skill saves you time, please give it a β β it helps other builders discover it!
When you hand Claude Code a raw idea ("I want to build X", "here's my startup idea"), it can write code immediately β but it usually skips the hard parts: validating the problem, finding the missing stakeholders, sequencing work so nothing is built before its foundation, and auditing for the security, privacy, and accessibility gaps that sink real products.
Ultimate Idea Watcher closes that gap. It behaves like a coordinated senior product organization β product strategy, design, engineering, security/privacy, and QA β fused into one skill that takes responsibility for the integrity of your idea across its entire lifecycle:
Raw Idea β Idea Preservation β Automatic Product Brief β Deep Research
β Problem Validation β User & Market Analysis β Product Direction
β Complete Product Architecture β Module & Workflow Architecture
β Phase Planning β Requirement Traceability
β Phase N Stage 1 Design β Design Verification & Gap Closure
β Phase N Stage 2 Implementation Plan β Implementation Verification
β Next Phase β Cross-Phase Audit β Production-Readiness Audit β Continuous Enhancement
You don't fill in a form. You drop in a sentence, rough notes, screenshots, or an existing codebase, and the skill self-initializes the brief, researches what matters, labels every assumption with a confidence rating, and keeps going.
- π§· Preserves your original idea verbatim while building a structured interpretation alongside it β never silently rewrites your intent.
- π Auto-fills a product brief (problem, users, features, business goal, security/privacy, devices, market, constraints) without making you complete it.
- π Separates the problem from the proposed solution so the first idea isn't blindly accepted as the best one.
- π Runs deep research with a source-priority model and converts findings into decisions β every major decision records evidence, confidence, alternatives, trade-offs, and the risk if wrong.
- ποΈ Defines the complete product before phasing: users & roles, product objects, modules, features, workflows, state machines, permissions, business model, and MVP boundary.
- π§± Splits work into dependency-safe phases (auth before protected workflows, canonical data before outputs, payments before paid marketplaces, audit logs before high-impact admin actions, β¦).
- π¨ Generates ready-to-paste prompts β a Stage 1 designer prompt (complete screen inventory, every state, responsive, accessibility, motion/3D with fallbacks, prototype wiring) and a Stage 2 engineering prompt (architecture, data, APIs, auth, security, jobs, testing, CI/CD, rollback).
- π‘οΈ Audits continuously: requirement traceability matrix, gap register with severity, contradiction detection, change-impact analysis, and a production-readiness audit.
- βΏ Treats security, privacy, and accessibility as defaults, not add-ons β it raises them before you ask.
| Field | Value |
|---|---|
| Skill name | ultimate-idea-watcher |
| Suggested command | /ultimate-idea-watcher |
| Type | Product discovery, research, architecture, planning, design, implementation & verification orchestrator |
| Operating mode | Self-initializing, research-driven, adaptive, phase-gated |
| Default output | Full Plan |
| Default design ambition | XHigh (Max) |
| Format | Single self-contained SKILL.md |
| License | MIT |
π Looking for a leaner, modular build with progressive-disclosure reference files and evals? See the companion repo ultimate-idea-watcher-v2.
Claude Code loads skills from ~/.claude/skills/ (personal, available in every project) or <project>/.claude/skills/ (project-scoped, shareable with your team).
git clone https://github.com/satishTheLegend/ultimate-idea-watcher.git \
~/.claude/skills/ultimate-idea-watcher# from your project root
git clone https://github.com/satishTheLegend/ultimate-idea-watcher.git \
.claude/skills/ultimate-idea-watchergit submodule add https://github.com/satishTheLegend/ultimate-idea-watcher.git \
.claude/skills/ultimate-idea-watcherAfter installing, restart Claude Code (or start a new session) so it discovers the skill. Verify with /skills or simply describe an idea and watch the skill activate. β
Claude Code reads the description in this skill's SKILL.md frontmatter and auto-activates the skill when your request matches β you don't have to invoke it manually (though /ultimate-idea-watcher works too).
It triggers when you:
- π‘ Share a new product, app, SaaS, platform, or marketplace idea β even a single line.
- πΊοΈ Ask to plan, scope, research, or validate a product.
- π§© Want work split into phases, a full screen inventory, a workflow map, or a design system.
- βοΈ Want design-before-implementation prompts for a designer or an engineer.
- π Ask to audit a plan for missing requirements, contradictions, or security/privacy/accessibility gaps.
- π Ask to verify or continue a previously planned product.
"I keep watching my parents struggle to run their small pharmacy β they track expiring medicines and reorders on paper and WhatsApp. Turn this into a real product plan."
β Preserves the raw idea, auto-fills a brief, researches the domain, separates the pharmacist (operator) from the owner (buyer), auto-includes security/privacy for sensitive data, produces a dependency-ordered phase plan + traceability matrix, and ends with an explicit next action.
"Here's my freelance-invoicing SaaS plan: sign up, create invoices, send to clients, pay via Stripe, dashboard for paid vs unpaid. What am I missing β especially security, privacy, and edge cases?"
β Surfaces concrete gaps (object-level access / IDOR, Stripe webhook verification, retention/deletion/consent, failed/partial payments, refunds, disputes, voiding invoices), assigns severity, and gives a remediation order β it won't rubber-stamp the plan.
"I'm building a habit-tracking mobile app β streaks, reminders, progress charts. Give me the complete screen list and a designer prompt for phase 1. Mobile-first and accessible."
β A Stage 1 deliverable: named screen inventory, every state (empty/loading/error/success), explicit accessibility, mobile as a real workflow, motion with reduced-motion fallbacks, prototype wiring with no dead controls.
ultimate-idea-watcher/
βββ SKILL.md # the complete skill β identity, engines, contracts, rules
βββ README.md # this file
βββ LICENSE # MIT
βββ .gitignore
This skill is intentionally a single, self-contained SKILL.md (the full operating manual β idea intake, research, architecture, phasing, Stage 1/Stage 2 contracts, verification, and 65 non-negotiable rules). The modular variant lives in the ultimate-idea-watcher-v2 repo.
The SKILL.md frontmatter follows the Claude Code skill format:
---
name: ultimate-idea-watcher
description: When this skill should be activated (idea β research β architecture β phases β prompts β audit)
---Found a gap, want to extend a contract, or improve an engine? Contributions are welcome! π
- π΄ Fork the repository.
- βοΈ Edit
SKILL.md(keep the frontmattername/descriptionintact and accurate β thedescriptionis what drives auto-activation). - π§ͺ Keep additions evidence-based, security/privacy/accessibility-aware, and dependency-safe.
- π Open a pull request describing the change and its rationale.
If Ultimate Idea Watcher helped you plan, scope, or ship something, please give it a star β it genuinely helps others discover the project and motivates continued work. π
π¬ Have an idea or found a bug? Open an issue or submit a PR.
Released under the MIT License β free to use, modify, and distribute.