The Crucible

The open standard for the agentic web.

The Foundry Standard defines what it means for a website to be AI agent-ready. Run The Assay to get your score. Earn your certification.

Three domains. One standard.

The Foundry Standard evaluates every site across three weighted domains. Each reflects a distinct aspect of AI agent compatibility.

Agent Accessibility30%

Can AI agents actually navigate and extract meaning from your site? This domain covers structured data, semantic markup, machine-readable metadata, and the presence of agent-friendly artifacts like agents.json and llms.txt.

  • Structured data (JSON-LD, schema.org)
  • Semantic HTML and ARIA roles
  • agents.json publication
  • llms.txt presence and format
  • Crawlability and robots.txt hygiene
Agent Security30%

Does your site handle AI agent interactions safely, without introducing risk to users or operators? This domain covers authentication patterns, rate limiting, input validation, and resistance to prompt injection vectors.

  • Authentication and authorization controls
  • Rate limiting and abuse prevention
  • Input sanitization and output encoding
  • Prompt injection surface area
  • HTTPS and transport security
Agent Safety40%

Does your site behave responsibly when AI agents act on its behalf? The highest-weighted domain covers consent surfaces, reversibility of agent-initiated actions, transparency disclosures, and human-in-the-loop checkpoints.

  • Consent and disclosure patterns
  • Reversibility of agent-triggered actions
  • Human-in-the-loop checkpoints
  • Transparency about AI involvement
  • Data minimization for agent interactions

Certification tiers

Earn certification by meeting the scoring thresholds across all three domains. Higher tiers require stronger fundamentals and active participation in the standard.

Bronzescore ≥ 60

Your site meets the baseline threshold for safe AI agent interaction.

  • Overall score of 60 or above
  • No critical failures in any domain
  • Passing marks in Agent Safety
Silverscore ≥ 75

Your site is actively signaling AI agent compatibility and meets strong standards.

  • Overall score of 75 or above
  • agents.json published and valid
  • No critical failures in any domain
  • Bronze requirements met
Goldscore ≥ 90

Your site exemplifies best-in-class AI agent readiness. The highest Foundry Standard certification.

  • Overall score of 90 or above
  • Annual certification audit
  • Silver requirements met
  • All three domains passing

Get Certified

Your site, certified agent-ready.

Certification isn't just a badge — it's proof that your site works correctly for the AI agents your customers are already using. Every certified domain appears in the public Foundry Standard registry, making your site discoverable to the tools and agents that power the modern web.

Real implementation, not instructions

We write the code, open the PR, and deliver working changes — not a report for you to act on.

Before & after score proof

We run The Assay before and after every engagement so you can see exactly what changed.

We do what we say

One-time pricing, no surprises. We deliver or we don’t charge — it’s that simple.

Don't have your score yet? Run The Assay first →

Bronze

“Send me the fixes”

$99one-time

no ongoing commitment

  • We implement all Assay report items
  • Delivered as files + step-by-step guide for your platform
  • You implement in under 30 minutes
  • One-time, no ongoing commitment
Most Popular
Silver

“Fix it for me”

$299one-time

no ongoing commitment

  • We attach to your repo
  • Open a branch with every fix implemented
  • You review and merge — done
  • One-time, no ongoing commitment
Best Value
Gold

“Keep me certified”

$599/yr

renews annually

  • Everything in Silver
  • Monthly Assay scans on your domain
  • Monthly status report emailed to you
  • We handle any score regressions
  • Renews annually