Ship Real MVP

What the company appears to be: Ship Real MVP is a self-reported project that describes itself as a pair of reusable, personalized Codex Skills for guiding AI-assisted software development. It aims to improve control and verifiability in building personal software from vague ideas and visual references.

What changed: The author reports having built two Codex Skills — one for risk-driven discovery and delivery, the other for UI reference calibration — and tested them against a baseline using GPT-5.6 Sol and Codex on an iOS camera app example.

Single most important open question: Does the described approach actually improve outcomes over default Codex capabilities in real-world use cases, or is it merely a conceptual framework without demonstrated traction?

This analysis is based entirely on the self-reported description provided by the author. No independent verification, revenue data, customer feedback, or performance metrics are available beyond what was stated.

The description states that Ship Real MVP consists of:

• Two reusable, personalized Codex Skills:
  • Ship Real MVP 02: Guides risk-driven discovery, narrow vertical-slice delivery, recovery states, verification, and a five-lens PM/UI/UX/Development/Testing review.
  • Adapt UI From References Mini 02: Separates visual references by layer ownership, rejects source-product semantics that should not transfer, calibrates one high-fidelity core state before expansion, and applies a Realness Gate against colored-wireframe results.

• A working SwiftUI camera app example named lilt Style Camera, which demonstrates the method in practice.

• The project is presented as both a tool (the reusable Skills) and an output (the iOS app).

The author claims these are not frameworks but methods for guiding Codex behavior. However, no evidence of actual usage beyond this single demonstration exists.

The author positions Ship Real MVP as:

• A way to make AI-driven software development more controlled and verifiable.
• An alternative to general-purpose tools or heavy frameworks.
• A method that improves fidelity to user intent when translating vague ideas into working code.

It evolved from the problem of:

• Default Codex inventing requirements not requested.
• Misinterpreting visual references.
• Lacking structured review processes.
• Not distinguishing between build and real product outcomes.

The author's claim is that personalized Codex Skills can help avoid these issues without adding complexity.

This is a self-described evolution from general AI capabilities to more guided, methodical use. There is no evidence of prior versions or market positioning beyond this one submission.

Not evidenced.

The description does not identify:

• Who uses the Skills.
• What types of users they target.
• Whether there are specific personas or industries involved.
• If the solution caters to developers, product managers, designers, or end-users.

No indication of a defined customer base or ideal customer profile exists in the provided text.

Not evidenced.

The description does not mention:

• How the Skills would be monetized.
• Whether they are sold as standalone tools, part of a platform, or offered via subscription.
• Any pricing model or revenue streams.
• If there is any commercialization plan beyond the hackathon submission.

No business model or pricing information is included in the self-report.

The project uses:

• Codex with GPT-5.6 Sol
• SwiftUI
• AVFoundation, Core Image, ffmpeg, GitHub, Pillow, Xcode

It includes:

• A working iOS camera app example.
• Installation instructions.
• A deterministic read-only audit prompt.
• Comparison reports and verification evidence.

The author tested both a baseline (general Codex) and skill-guided version using the same inputs but different approaches.

The technical stack is detailed, but no evidence of scalability, infrastructure, or delivery beyond this one example exists.

Not evidenced.

There is no mention of:

• Customers or users.
• Revenue or monetization.
• Product adoption or usage metrics.
• Iteration history or roadmap.
• Any form of traction beyond the single hackathon submission.

No signs of traction, growth, or maturity are evident in the description.

Not evidenced.

The description does not:

• Identify competitors.
• Compare to existing tools or platforms.
• Describe how it fits into the broader AI development ecosystem.
• Mention any competitive advantages or differentiation strategies.

No competitive landscape or positioning relative to others is described.

Inferences based on self-report:

1. Unproven effectiveness: The author notes that "there was no universal winner" and that default Codex was sometimes faster, suggesting the Skills may not consistently improve outcomes.
2. Limited scope: Only one developer (hua hao) is involved; no team or organizational structure is described.
3. No commercial viability: No evidence of monetization, market demand, or product-market fit beyond a hackathon project.
4. Unclear scalability: The Skills are described as reusable but not demonstrated at scale or in other domains.
5. Self-reported only: All claims are unverified and lack external corroboration.

These risks stem from the lack of independent validation and real-world application data.

1. What specific problems do you see users facing when using default Codex, and how does your solution address those?
2. How many developers or teams have tested these Skills in practice?
3. Have you run any A/B tests comparing the personalized Skills to baseline Codex usage?
4. Is there a plan for scaling beyond this single example?
5. What is the intended business model for monetizing these Skills?
6. Can you demonstrate how the Skills integrate with other development workflows or tools?

These questions aim to uncover whether the described solution has real-world utility and commercial potential.