DW–001 / Offers MCP / Live / v0

Offers MCP.

Deep purchase intelligence for agents, applications, and humans. A single tool to scan many stores and return live prices, true cost with shipping, delivery details, and merchant trust scores.

Try it now GitHub ↗

Free. No signup. No API key. Early v0, and we want your feedback.

Quickstart

Try it now.

One command connects it to Claude. Or hit the plain HTTP endpoint and see the raw answer. No account, no key.

Claude Code

claude mcp add --transport http offers https://offers-engine.dexter-works.com/mcp

ChatGPT, Claude Desktop, Cursor, VS Code step-by-step setup in the README ↗.

Then just ask

"Use Offers MCP to check if I'm getting a good deal on the Sony WH-1000XM6 at $398. I'm in 83702."

Or plain HTTP

curl "https://offers-engine.dexter-works.com/api/search_offers?q=Sony+WH-1000XM6+black&zip=83702"

Why this

Purchase Intelligence
beats AI Search

Claude and ChatGPT answer shopping questions from search-engine snippets: cached previews, prose summaries, and tell you to "check the store to be sure." Offers MCP checks merchants live and delivers decision-ready data.

	Built-in AI search	Offers MCP
Prices	Stale search snippets	Live offers, timestamped at request time
Delivery Timeline	Unavailable	Delivery windows by zip code
Application support	Requires custom tooling and permissions.	Structured JSON contract with confidence ratings
Real Cost Calculation	Excludes shipping	Landed cost: price + shipping
Trust	Whatever ranks in search	Merchant trust scores; untrusted listings filtered
Agent effort	Several searches, page reads, and reasoning per question	One tool call, one structured answer
User data	Can be leveraged for personalization	No per-user tracking: queries stay private, never sold, never used for retargeting

How it works

Shop
everywhere
in seconds.

The deal exists. Finding it is the tedious part. Offers MCP does the cross-merchant research and hands your application structured data to build on.

01 / Robust

Built to be trusted

Live prices across merchants, shipping folded into a true landed cost, untrusted listings filtered, every merchant rated with a trust score. Every claim carries a confidence score and a timestamp. When it isn't sure, it says so instead of guessing.

02 / Private

Privacy First

No per-user tracking, no raw IPs stored. The comparison happens on our servers, and a merchant learns about a shopper only at click-through to buy. Queries stay private: never sold, never shared, never used for retargeting.

03 / Efficient

Deep and fast research

Free text ("Sony WH-1000XM6 black") or a barcode number in: one call fans out to every source in parallel and returns one structured answer: the best trusted deal, plus the full option list.

Under the hood

Quality Data.

The short version of how we deliver real-time results, so you, or your AI agent, can make the best decision with all the options in view.

Partner relationships and structured data. We work through official merchant and catalog partner programs, structured data sources, and creative insights to stay accurate, efficient, and private.
Parallel retrieval, one schema. A single input fans out to many sources at once; everything comes back normalized in a structured contract you can build against.
Provenance on every offer. Each result names its source and the moment it was checked. Where sources disagree or fall short, the offers mcp shares insight.
Where it fits: cross-vendor price checks from the store aisle, or millions of programmatic lookups inside your application. Same tool call, same answers.

Early access

Test it. Break it.
Tell us.

This is v0: US consumer goods, a small set of sources done carefully. Real queries from real shoppers are exactly what makes it better.

Report a result ↗ hello@dexter-works.com

Got a wrong price, a bad match, or a store we should trust (or shouldn't)? Send the exact query. That's the gold.