Report card · Coding agents
Factory
93/100. Top of the class. Practices what it sells.
Scanned https://factory.ai/ on October 30, 2026.
Subjects
- Arobots.txt rules for AI agentsGood work.All 11 AI agents checked may fetch every key page. robots.txt names 8 AI agent(s) explicitly.
- Bllms.txt and llms-full.txtGood work./llms.txt found (51858 chars, 198 link(s)). No /llms-full.txt.Extra credit: Improve llms.txt: publish /llms-full.txt with the full docs inlined.
- ASitemapGood work.Sitemap at https://factory.com/sitemap.xml (from robots.txt) with 424 entries.
- AServer-rendered readable HTMLGood work.All 1 key page(s) return readable HTML without running JavaScript.
- BMarkdown versions of key pagesGood work.Every key page has a markdown version.Extra credit: Advertise the markdown versions of / with <link rel="alternate" type="text/markdown">.
- AStructured data (JSON-LD)Good work.Every key page has valid JSON-LD.
- AOpenAPI and API docs discoverabilityGood work.Machine-readable API spec at /openapi.json.Extra credit: Add /.well-known/api-catalog (RFC 9727) pointing at the spec so agents can find every API from one URL.
- CMCP server discoverabilityMentions MCP, never shows up with it.An MCP endpoint is documented (https://docs.factory.com/harness/mcp), but no /.well-known MCP discovery document.Extra credit: Publish /.well-known/mcp.json describing the server. Discovery through /.well-known/mcp.json is still a draft proposal, so also document the server URL, transport, and auth in llms.txt and your docs.
- AAuth, signup, and bot-challenge frictionGood work.No login walls, bot challenges, or CAPTCHAs on 1 key page(s).
- APage weight and token costGood work.Key pages are light enough to read in one pass.
Think the grade is wrong? Fix the site and ask for a regrade, or run the scanner yourself. Scores are reproducible. How grading works.