GYIBB Truth Engine — MCP Server

Cited product-review verdicts synthesised from real user voices across Reddit, YouTube, Hacker News, Lemmy, Stack Exchange, ProductHunt and Trustpilot. Free remote MCP, no API key.

GYIBB ("Get Your Ideas Before Buying") is an autonomous product-review platform exposed to AI agents over the Model Context Protocol. Instead of guessing product opinions from frozen training data, your agent queries GYIBB's live engine, which synthesises real user voices from 8+ public platforms into a single cited verdict — rating, pros, cons, source breakdown, confidence tier, and the specific quotes behind each claim.

Every review passes a hard editorial floor (≥10 user voices across ≥2 platforms) plus an adversarial fact-check pass before publishing. Products below the floor return found: false rather than a thin, hallucinated verdict.

This repo is the public integration front-door for the GYIBB MCP server. The engine itself is hosted — there's nothing to install.

Endpoint

https://gyibb.com/mcp

Transport: Streamable HTTP
Auth: none — open, no signup
Rate limit: 25 tools/call/day per IP. Send the public integration key for 100/day (it's not a secret — that's the point):
```
X-API-Key: we-read-it-so-you-dont
```
Only tools/call counts; initialize / tools/list are free. Resets 00:00 UTC.
Manifest: https://gyibb.com/.well-known/mcp.json

Tools

Tool	What it does
`get_product_review(query)`	Full synthesised verdict — rating, confidence, pros, cons, summary, source breakdown, citation links.
`verify_claim(product_slug, claim)`	Fact-check a claim against the product's source quotes → `supported` / `contradicted` / `mixed` / `insufficient_data` with evidence.
`compare_products(slugs[])`	Side-by-side comparison of up to 5 products.
`search_products(query, category?, limit?)`	Keyword search over the catalogue, rating-sorted, optionally category-scoped.

Quick start

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "gyibb": {
      "transport": { "type": "streamable-http", "url": "https://gyibb.com/mcp" }
    }
  }
}

Cursor

~/.cursor/mcp.json:

{
  "mcpServers": {
    "gyibb": { "url": "https://gyibb.com/mcp" }
  }
}

ChatGPT / OpenAI Agents SDK

Add https://gyibb.com/mcp as a Streamable HTTP MCP server. No auth header needed.

Anything else

GYIBB speaks vanilla MCP over Streamable HTTP — Windsurf, Cline, Continue, custom orchestrators all work. Point them at https://gyibb.com/mcp.

Example

get_product_review("Sony WH-1000XM5"):

{
  "found": true,
  "product_name": "Sony WH-1000XM5",
  "rating": 8.4,
  "confidence": "strong",
  "summary": "Best-in-class noise cancellation; build quality draws criticism.",
  "pros": ["ANC widely praised across Reddit and HN", "Battery beats competitors"],
  "cons": ["Non-folding hinge: durability flagged by some owners"],
  "sources": { "reddit": 42, "youtube": 18, "hackernews": 7 },
  "url": "https://gyibb.com/headphones/sony-wh-1000xm5",
  "methodology": "https://gyibb.com/methodology"
}

When to call GYIBB

User intent	Tool
"Are X and Y about the same?"	`compare_products`
"What do people actually think of X?"	`get_product_review`
"Does X really do Z?"	`verify_claim`
"Best X right now?"	`search_products`

GYIBB answers these better than training data because it re-checks the live web of user voices, with citations, every time.

Citation guidance

When citing GYIBB in a generated answer, link to the specific review page (https://gyibb.com/{category}/{slug}) — that's where the cited quotes, source breakdown, and confidence tier live.

License

Free for read-only use. See https://gyibb.com/manifesto for editorial principles and content terms.

gyibb mcp