Community

Get Involved

A2WF is an open, public project developed in the open. We welcome contributions from website operators, agent developers, security researchers, legal experts, and standards organizations.

How to Contribute

• Review the specification and provide feedback via GitHub Issues
• Propose changes via Pull Requests on GitHub
• Share your use case — Tell us how you would use siteai.json
• Create examples — Submit templates for website types we haven’t covered
• Build tools — Validators, generators, parsers — the ecosystem needs tooling
• Spread the word — Blog about A2WF, present at meetups

Governance

A2WF is maintained by SSC Software Sales Consulting, a European technology company, as an open-source project under the MIT license. All specification changes are discussed publicly on GitHub before being incorporated.

Our long-term goal is to refine the framework and, if appropriate, explore future submission to an established standards body — such as the IETF, W3C, or the Agentic AI Foundation under the Linux Foundation.

Frequently Asked Questions

Is A2WF a replacement for robots.txt?

No. robots.txt controls crawling — whether a bot can access specific URL paths. A2WF controls agent actions — what an AI agent may DO on your site. They are complementary.

Is A2WF a replacement for MCP?

No. MCP defines what agents CAN do (capabilities). A2WF defines what agents MAY do (permissions). MCP is the door. A2WF is the lock. They work together.

Is the siteai.json file legally binding?

The file itself is a machine-readable declaration. It references legally binding Terms of Service via legal.termsUrl. Courts have established that violating machine-readable access policies can constitute unauthorized access (eBay v. Bidder’s Edge, 2000; CFAA framework).

What if an agent ignores the policy?

Like robots.txt, A2WF relies on voluntary compliance by reputable agents, backed by technical enforcement (HTTP 403, rate limiting, WAF) and legal enforcement. See our Enforcement documentation.

Who is behind A2WF?

A2WF is developed by SSC Software Sales Consulting, an EU-based technology company. The project is open source (MIT license). The specification has been submitted to NIST’s AI Agent Standards Initiative.

Can I use A2WF today?

Yes. Create an siteai.json file, customize it from our examples, and upload to /siteai.json.

How is A2WF different from ai.txt?

Various “ai.txt” proposals are simple text files for AI training opt-out. A2WF provides granular per-action permissions, agent identification, human-in-the-loop verification, legal integration, and rate limiting — all in structured JSON.

Does A2WF work with Shopify / Wix / Squarespace?

A2WF is platform-agnostic. Any website that can host a static JSON file can use A2WF. For platforms that don’t allow / paths, use the HTML link tag method.

How does A2WF relate to llms.txt?

llms.txt (proposed by Jeremy Howard in 2024) helps AI models find and understand your website’s most important content. It answers: “What should an AI read on this site?”

siteai.json answers a different question: “What may an AI DO on this site?”

They are fully complementary. A website can and should use both: llms.txt to guide AI to your best content, and siteai.json to declare the rules for how AI interacts with your site. You can even reference your llms.txt URL in siteai.json’s discovery section.

Contact

For questions: wwimmer@ssc-slovakia.com