We open-source our agentic-commerce tooling.
Anyone can claim “agent-ready” or “secure.” We publish the tools that prove it: MIT-licensed, with honest limitations in every README and benchmarks you can reproduce. Two are live, two are on the way.
MIT-licensed · deterministic · reproducible benchmarks
- LiveagentreadyIs your store visible to AI shopping agents?
- LivevibecheckIs your AI-built store secure & production-ready?
- October 2026retail-mcpExpose your catalog to AI agents, securely.
- October 2026acp-testkitTest your agentic checkout before a real agent does.
Four tools. One honest program.
Each maps to a real problem we fix for clients, and each ships with the MnT Future repo standard: a 30-second quickstart, honest limitations, and tests you can run.
retail-mcp
Expose your catalog to AI agents, securely.
Open-source Retail MCP server kit: catalog, inventory and pricing for AI agents. Headless engines first; Shopify already ships its own.
acp-testkit
Test your agentic checkout before a real agent does.
Agentic-checkout conformance suite + a mock shopping agent for your ACP / Google UCP integration.
Follow along: github.com/MnT-Future · or get release notes by email
Senior by default. Honesty over hype.
The same standard runs through every repo, so a tool of ours reads like our engineering, not a demo.
Prove it, don't claim it
The agentic-commerce protocols are months old and full of bold claims. We publish working, inspectable tools instead: run them yourself, no marketing required.
Honest limitations
Every README documents what the tool can't do. Heuristics are labelled, benchmarks state their conditions, and evidence is quoted from your own code or store.
MIT, no lock-in
Use it, fork it, build on the engine. The code is free forever; the engineering behind the fixes is our business, not the tool.
Maps to real work
Every tool ladders into a service we actually deliver for clients. No orphan repos, no abandonware: each one earns its place.
Fair questions, honest answers.
Yes, MIT-licensed, forever. The CLIs are free, agentready's hosted scan is free, and no email is needed for a score. Our business is the engineering behind the fixes, not the tools.
vibecheck's CLI scans your code entirely on your machine: nothing reaches us. agentready's CLI fetches your store directly from your machine. Only agentready's web scanner has our server fetch your public pages to compute the score shown to you; we keep your email and grade only if you request the fix plan.
Anyone can propose a change (fork → pull request), but nothing merges without our review, passing tests, and a signed-off commit. Only the MnT Future team can push to the main branch.
The agentic-commerce and AI-codegen waves are new and full of bold claims. Publishing working, inspectable tools is how we prove we're practitioners, and teams who discover gaps with our tools often ask us to fix them. Honest exchange.
Our tools find the gaps. We build the fixes.
Run any of them, then bring the report to a free strategy session: a senior consultant maps every finding to a concrete fix plan for your platform.