/idea-validator

name: "idea-validator"
description: "Validate app/startup ideas with market, feasibility, commercial, and open-source competitor analysis. Use when asked to evaluate, validate, or score a product idea. Don't use for PRDs, go-to-market plans, or investor decks."
license: MIT
effort: max
metadata:
  version: 1.6.0
  author: "Luong NGUYEN <luongnv89@gmail.com>"

Idea Validator

Critically evaluate ideas with honest feedback on market viability, technical feasibility, and actionable improvements.

When to Use

Trigger this skill when the user asks to:

• Evaluate, validate, or score an app idea or startup concept
• Get honest feedback on whether an idea is worth building
• Research what competitors already exist in a space, including commercial tools/services and open-source alternatives
• Turn a vague concept into a structured validation report

Instructions

Run the steps in this order: Setup → Phase 1 Clarify → Phase 2 Technical Context → Phase 3 Competitive Landscape → Phase 4 Evaluate → Phase 5 Improve → Commit and Push → Final Report. Do not skip or reorder a step, except where an Edge Cases row says to stop. A stop after Setup step 4 still runs Commit and Push on the files written so far, then the Final Report. If the ideas root is the root of an `ideas` repo, also run README Maintenance after each file update (detection rule in that section).

Terms used throughout:

• **Ideas root** — the directory that holds every idea folder (resolved in Setup step 2).
• **Project folder** — `<ideas root>/YYYY_MM_DD_<short_snake_case_name>/`, holding this idea's `idea.md` and `validate.md`.
• **Status** — the run outcome, `COMPLETE`, `PARTIAL` or `BLOCKED`, chosen by the rules in Final Report.

Repo Sync Before Edits (mandatory)

Run this in the ideas root, before Setup step 3 creates or changes any file. Skip it when the ideas root is not inside a git repository (`git -C <ideas root> rev-parse --git-dir` fails), and record that skip for the Final Report.

If `git status --porcelain` prints nothing, sync directly:

cd "<ideas root>"
branch="$(git rev-parse --abbrev-ref HEAD)"
git fetch origin
git pull --rebase origin "$branch"

If it prints anything, stash first, sync, then restore:

git stash push -u -m "pre-sync"
branch="$(git rev-parse --abbrev-ref HEAD)"
git fetch origin && git pull --rebase origin "$branch"
git stash pop

If `origin` is missing, the pull fails, or the rebase or stash pop conflicts, stop and ask the user before continuing.

Setup

1. **Get the idea.** Use the idea in `$ARGUMENTS`. If `$ARGUMENTS` is empty, ask the user to describe the concept. If no description arrives, stop with status `BLOCKED`. 2. **Resolve the ideas root** (tool-agnostic; ask once per environment). Take the first that applies: 1. The env var `IDEAS_ROOT`, if set. 2. The path in `~/.config/ideas-root.txt`, if that file exists. 3. The path in legacy `~/.openclaw/ideas-root.txt`, if that file exists. Copy the value into `~/.config/ideas-root.txt`. 4. Otherwise, ask the user where to store generated docs. Suggest `~/workspace/ideas`. Save the answer to `~/.config/ideas-root.txt`.

Ask again only when the user asks to change the location. 3. **Create the project folder** under the ideas root. If a folder with the same date and name already exists, reuse it and update its files instead of creating a second one. 4. **Create `idea.md`** with the idea and the clarifications. 5. **Create `validate.md`** for the evaluation and recommendations. 6. **Echo the absolute project folder path** in your response, so downstream skills can pick it up.

If the ideas root cannot be created or written, stop with status `BLOCKED` and name the path.

Read `references/file-templates.md` when creating either file or updating a section named in Phases 1-5. That file owns header names and order; the phase instructions own the content.

Phase 1: Clarify the Idea

Ask the user these questions with the question tool (`AskUserQuestion` or equivalent). If no question tool exists, ask in plain chat. Skip a question that the idea description from Setup step 1 already answers.

• What problem does this solve? Who has this pain?
• Who is your target user? Be specific.
• What makes this different from existing solutions?
• What does success look like in 6-12 months?

Update `idea.md` with the responses. Write `unknown` for each question the user declines or cannot answer, and list it under Uncertainty in the Final Report. The target-user question is the exception: see Edge Cases.

Phase 2: Gather Technical Context

Ask the user, with the same tool and skip rule as Phase 1:

• Preferred tech stack?
• Timeline and team size?
• Budget situation (bootstrapped/funded/side project)?
• Existing assets (code, designs, research)?

Update the `idea.md` Technical Context section. Write `unknown` for each unanswered item.

Phase 3: Competitive Landscape Research

Before evaluating the idea, perform live web research to find what already exists in the space. Do not rely on memory or training data for market, pricing, traction, competitor, or open-source claims — web search is mandatory so the report reflects current information.

Use the available web search tool (`WebSearch`, `web_search`, or equivalent) to run at least 4 varied queries covering:

**Commercial tools/services** — SaaS products, mobile apps, enterprise platforms, paid APIs, agencies, marketplaces, and other commercial offerings solving the same or adjacent problem. Search the core problem statement plus keywords like "app", "tool", "platform", "SaaS", "startup", "pricing", "alternative", and audience-specific terms.

**Open-source solutions** — GitHub/GitLab repositories, self-hosted tools, packages, frameworks, templates, and OSS alternatives that solve the same problem or provide a strong foundation. Search with terms like "open source", "GitHub", "self-hosted", "OSS", "alternative", "library", and relevant package registry n