AI Agents

37 / 100

How well does your parser actually work? Evaluate document parsing across tables, charts, content faithfulness, semantic formatting, and visual grounding.

Scanned 2h ago · 6 views · Open site

Checks

• HTTPS5/5

  Served over HTTPS.

• llms.txt12/12

  Found /llms.txt (2843 chars).

• Agent manifest0/15

  No /.well-known/ai-agent.json.

  Fix: Publish https://parsebench.ai/.well-known/ai-agent.json with name, description, capabilities, contact, auth, and pricing.

• A2A agent card0/12

  No /.well-known/agent-card.json.

  Fix: Publish https://parsebench.ai/.well-known/agent-card.json with name, url, and skills or capabilities.

• MCP endpoint12/12

  Discoverable MCP metadata or endpoint.

• OpenAPI spec8/12

  An OpenAPI-like file was found, but it did not fully parse as a spec.

  Fix: Serve /openapi.json (or /swagger.json) with a valid OpenAPI spec, or link it from the homepage.

• Named capabilities0/8

  No named capabilities, skills, or API operations.

  Fix: List named capabilities in the agent manifest, agent card skills, or OpenAPI path summaries.

• I/O schemas0/8

  No I/O schemas.

  Fix: Add components.schemas / requestBody schemas in OpenAPI, or input/output schemas on each skill.

• Machine-readable pricing0/8

  No machine-readable pricing.

  Fix: Add a pricing object to ai-agent.json, or publish /pricing.json with currency and product amounts.

• Auth, contact, rate limits0/8

  No contact, auth, or rate limit declaration.

  Fix: Declare contact, auth, and rateLimit in ai-agent.json (and OpenAPI info.contact / securitySchemes).

7 checks short of a perfect 100. The fix lines above are the report — what to publish, at which path.

How scoring works · Recently graded · Back to the board