Ora
Agent-readiness scanner from era labs: scores how well AI agents can find, read and use your website, runs real agents against it and suggests fixes
Overview
Ora is an agent-readiness scanner made by era labs. You give it a URL and it scores, on a 0-100 scale with a letter grade, how well AI agents can find your product, read it, sign in and actually use it. Its Deep Scan does not only check that files such as llms.txt exist: it sends real agents and crawler identities (ChatGPT, Claude, Gemini, Perplexity, OpenClaw and Ora's own agent) at the site, tries live OAuth and MCP handshakes, and runs multi-turn tasks such as booking a demo. The result is a scorecard with the checks that failed and what to fix. Ora also offers Journey runs that record where an agent stalls on your site, category leaderboards, a WebMCP audit, a public REST API and the ax CLI for scoring a URL from a terminal or a coding agent. The company says it was founded by people who worked with Anthropic and OpenAI on agent standards, and it publishes the scoring rules as an open standard called AgentReady.
The Verdict
Who Should Use Ora?
Best For
- Teams that want AI assistants and agents to find, recommend and use their product
- Developers shipping an MCP server, OAuth flow or llms.txt who want a live test, not a checklist
- Site owners who want a repeatable score to track agent-readiness over time
- Coding-agent workflows that can call the
axCLI or the API to audit a URL
Not Ideal For
- Classic SEO or accessibility audits for human visitors (use a dedicated SEO tool or axe-core)
- Teams that need a vendor with public pricing, SLAs and a named company profile today
- Products that have no reason to be used by agents
What's Great
- Runs real agents and live protocol handshakes instead of only checking that files exist
- Free, public scan that takes about a minute
- Scorecard marks checks that could not run as such, rather than passing or failing them
- Read API endpoints need no key, and the
axCLI needs no account - Published methodology and an open AgentReady standard, so the score is explainable
- Large public dataset: its 2026 report scored more than 140,000 products, with a median of 44 out of 100
Watch Out For
- No pricing page; the site does not say whether paid plans will follow
- The scoring model is still moving: the Deep Scan post describes five layers, while the current site and agent card describe four (discovery, access, usability, payments)
- Scans make hundreds of LLM calls, so results can shift by a couple of points between runs
- Little public company detail: the about page does not name founders, location or founding year
- No independent user reviews found on G2, Reddit or Hacker News
Pricing
View all features & details
Key Features
- Scan: 0-100 score and letter grade for any URL
- Journey: point an agent at a domain with an intent and see where it succeeds or stalls
- Leaderboards by category (consumer, e-commerce, travel, CRM and more)
- WebMCP audit at webmcp.ora.ai
- Directory of scored products and published research
- REST API for scans, scores and checks; write calls verify the caller is an agent
axCLI for audits from a terminal or coding agent
What Deep Scan Tests
- Discovery: does the product surface in AI answers, and are well-known files in place
- Identity: does an LLM's summary of the product match what it does
- Auth and access: live OAuth with PKCE, scopes, refresh and Web Bot Auth
- Integration: live MCP sessions, x402 payment handshakes and A2A agent cards where advertised
- User experience: goal success on realistic multi-turn tasks
Use Cases
- Checking a site before and after adding llms.txt, an MCP server or agent sign-in
- Benchmarking your product against others in its category
- Letting a coding agent audit and fix a site in one loop
Sources: Ora · Methodology · Introducing Deep Scan (April 2026)
How It Compares
| Feature | Ora | axe-core | Manual llms.txt check |
|---|---|---|---|
| Audience tested | AI agents and crawlers | Human users with assistive tech | AI crawlers |
| Method | Live agent runs and protocol handshakes | Rule-based page tests | File exists or not |
| Score | 0-100 with letter grade | Pass or fail per rule | None |
| CLI | Yes (ax) | Yes | No |
| Price | Free scan | Free, open source | Free |
Sources: Introducing Deep Scan · axe-core
Is this your tool? Add a badge