vibe-check Product Discovery Skill
Turns a complete beginner's vague app idea into a validated, buildable plan — before any code gets written.
Pure instruction-based skill: no malware, no credential collection; version check is non-blocking best-effort with silent skip on failure; explicitly forbids fabricated quotes, intercepts harmful ideas; MIT license with clear attribution. Deductions: runtime dependence on external fetching (Reddit, review sites, raw.githubusercontent.com), data flowing through unverified mirror instances, no explicit privacy/data-flow disclosure.
Highly self-consistent instructions: three-rung fetch ladder, fail-closed source verification, mermaid/table fallbacks when the diagram engine cannot render, and documented behavior on abnormal inputs. Deductions: static review cannot execute; no test suite or CI evidence covering the skill's key paths is present in the provided files, so the reliability ceiling of 10 applies.
Trigger conditions, three on-ramps, and non-fit boundaries (not for experienced dev debugging/CI) are exceptionally clear in the frontmatter, supporting accurate semantic invocation. Deductions: no Chinese-language support declared; core discovery depends on Reddit, G2, GitHub — services unreachable from mainland China — with only partial mitigations (Serper, hand-off to user).
Semantic versioning, a detailed CHANGELOG including self-reported defects, bump.sh with thorough pre-flight checks, RELEASING.md maintenance path, well-layered progressive disclosure across reference files, explicit license. Deductions: unverified solo maintainer with no third-party governance; self-documented fragility of the hardcoded master-branch version URL.
Claimed deliverables (structured plan plus self-contained interactive PRD) are backed by referenced examples and a clear value proposition over manual planning. Deductions: static review cannot verify representative output quality; SKILL.md is very long, implying substantial context cost; example files are not included in the provided evidence.
The CHANGELOG is an auditable evolution record including self-reported bugs and fixes; README cites real products (clearlist.me) and methodological sources; fact/inference separation is decent. Deductions: key artifacts (PRD, validation report, examples) are not within the provided files for cross-checking; execution claims (e.g., headless-browser render fix) rest on author assertion only.
- Discovery depends on Reddit and overseas review sites, largely unreachable from mainland-China networks; expect degraded results and reliance on the fallback ladder or manual user search.
- Runtime version checks and discovery involve outbound requests to third-party sites; mirror instances are unverified — assess in sensitive environments.
- This is a static review with no execution; discovery and output quality depend on model execution — trial on a small task first.
- Publisher is an unverified solo maintainer with no third-party maintenance guarantee.
What does this skill do, and when should you use it?
vibe-check is an open-source skill for AI coding tools, written by a product manager with 12+ years of 0-to-1 experience. It targets people who have never coded: it first digs out the real problem behind an idea through questioning and Reddit/review-site research, pressure-tests whether it's worth building using ODI opportunity scoring, then produces a full experience blueprint, tech stack, and build plan. It also teaches build-time basics like Git and deployment, and offers a Checkup Mode to untangle AI-messy codebases. Three on-ramps cover the full journey, validate-only, and plan-only paths.
Runs a two-stage flow: first calibrates the user's experience level and routes to the right on-ramp; Phase 0 grills hypotheses out of the user, then mines Reddit and review sites (G2, Capterra, etc.) for verbatim user quotes, scores each need with Pain + max(0, Pain − Served), and produces an Opportunity Map and Competitor Matrix. It then runs Crazy 8 experience sketches, a Story Map, user flows, tech stack recommendations, cost estimates, and growth-loop design, finishing with a Markdown build plan plus an interactive browser-opened PRD. Checkup Mode reads references/CODE-CHECKUP.md to safely tidy a messy codebase. All boards render through a bundled diagram engine with mermaid fallback.
- A non-coder with an app idea and no idea where to start, who needs to walk from concept to a buildable plan.
- A vibe coder who can get something on screen but needs help thinking through the full product picture before building.
- Someone who only wants a straight, evidence-backed verdict on whether their idea is worth building — a findings summary, not a blueprint.
- A person who already validated their idea and brings real research, skipping discovery and jumping straight to MVP planning.
- A builder whose AI-built app became messy and scary to touch, needing Checkup Mode to safely organize it.
- A beginner mid-build who needs plain-language explanations of GitHub accounts, commit/push, deployment, or keeping API keys safe.
What are this skill's strengths and limitations?
- Fills a real gap: most skills assume you already know what to build; this one handles everything before that decision.
- Grounded methodology: combines Jobs-to-be-Done, ODI opportunity scoring, Design Sprint (Crazy 8, future press release), and user story mapping.
- Exceptionally beginner-friendly: one question at a time, every term explained in plain language, a running "Words You Now Know" glossary.
- Strict evidence discipline: quotes are re-verified against permalinks, unverified material is dropped or flagged, fabrication is explicitly forbidden.
- Backed by real output samples: ClearList is a real, live product built with the skill.
- MIT licensed, prompt-only (no scripts), easy to audit and modify.
- Discovery depends on fetching Reddit and review sites, which many AI tools cannot do; a fallback ladder exists but result quality varies by environment.
- Research is directional, not statistical (Reddit and reviews stand in for hundreds of customer interviews), so conclusions deserve caution.
- Explicitly not for experienced developers debugging, refactoring, or setting up CI/CD, and not for non-app projects.
- No automated test suite; effectiveness rests entirely on the underlying model's ability to follow long instructions.
- SKILL.md was shown truncated; details after Phase 2 could not be fully verified.
How do you install this skill?
Easiest via the skills CLI: run npx skills add TexasBedouin/vibe-check. Or clone into your project's .claude/skills directory: git clone https://github.com/TexasBedouin/vibe-check .claude/skills/vibe-check. To update: npx skills update (CLI install) or git pull inside the skill folder. For other AI tools, paste SKILL.md contents into your system prompt or project instructions.
How do you use this skill?
After installing, tell Claude: "Use the vibe-check skill to help me plan my app." Or trigger a specific on-ramp: "Is my idea worth building? Reality-check it with vibe-check." or "I already validated my idea. Use vibe-check to plan the build." The skill runs a non-blocking version check at session start and begins with a light calibration question.
How does this skill compare with similar options?
The author explicitly differentiates from Matt Pocock's grill-me: grill-me is relentless questioning for engineers, vibe-check is for everyone else. It also adapts improve-codebase-architecture (as Checkup Mode), autoresearch (the verify-and-iterate loop), and Garry Tan's /office-hours (problem reframing).