name	onesvibe-prior-art
description	Prior-art check against 2,400+ live, machine-verified vibe-coded apps. Use when the user proposes building something ("I want to make X", "has anyone built X?"), or wants to browse/filter AI-built products by category, tag, or tech stack.

One's Vibe prior-art check

One's Vibe is a curated gallery of AI-built products that are live and testable — every entry's URL is re-checked on a rolling ~3-day cycle. Before the user builds something, check whether someone already vibe-coded it; finding nothing similar is also a useful answer.

Which source to use

Default to the GitHub snapshot — it is a static CDN file and costs the backend nothing. Only call the live API when you need semantic similarity.

1. Browse / filter / analyze → GitHub snapshot (preferred)

Fetch once, then filter locally:

https://raw.githubusercontent.com/JefferyLee/awesome-vibe-coded-apps/main/data/projects.json

One JSON array, regenerated daily, CC0. Fields per app: slug, title, short_description (+ short_description_zh), live_url, primary_category (one of: games_play, productivity, education, creative_tools, ai_agents, developer_tools, finance_business, health_wellness, social_community, utilities, other), tag_slugs, detected_stack (e.g. "Next.js", "Supabase"), is_featured, github_stars, recent_try_count, published_at, live_check_status.

This answers keyword matching, "list all AI writing tools", "what's built with Next.js", popularity sorting, and stats — without any API call. data/graveyard.json in the same repo lists apps that have since gone dark.

2. Semantic similarity ("has someone built this idea?") → live API

Only when keyword filtering isn't enough and you need meaning-level matches:

GET https://onesvibe.app/api/reference?q=<the idea, plain words>&limit=6

No auth. Returns matches with provenance (how each was found and when it was last checked). There is also an MCP server exposing the same search as a tool: claude mcp add --transport http onesvibe https://onesvibe.app/api/mcp.

Reporting results

Give each match's name, one-line description, live URL, and its detail page https://onesvibe.app/projects/<slug>.
Say what's different about the user's idea versus the matches, not just that matches exist.
"Live checked" means the URL was recently reachable. It is not a safety audit, quality endorsement, or popularity claim — never present it as one.

One's Vibe prior-art check

Which source to use

Default to the GitHub snapshot — it is a static CDN file and costs the backend nothing. Only call the live API when you need semantic similarity.

1. Browse / filter / analyze → GitHub snapshot (preferred)

Fetch once, then filter locally:

https://raw.githubusercontent.com/JefferyLee/awesome-vibe-coded-apps/main/data/projects.json

2. Semantic similarity ("has someone built this idea?") → live API

Only when keyword filtering isn't enough and you need meaning-level matches:

GET https://onesvibe.app/api/reference?q=<the idea, plain words>&limit=6

Reporting results

Give each match's name, one-line description, live URL, and its detail page https://onesvibe.app/projects/<slug>.

Say what's different about the user's idea versus the matches, not just that matches exist.

"Live checked" means the URL was recently reachable. It is not a safety audit, quality endorsement, or popularity claim — never present it as one.

onesvibe-prior-art