generated: '2026-08-13' method: probed source: https://mcp.b12.io/mcp + https://www.b12.io/pricing/ limit_count: 0 note: >- B12 publishes NO HTTP rate limit for its MCP surface and returns no rate-limit headers. limit_count is an honest zero: it counts published request-rate limits, and there are none. What B12 does publish is a plan-attached AI credit quota, recorded below as consumption quotas — those govern the product, not the endpoint, and an agent cannot read them at runtime. limits: [] response_headers: observed: [] x_ratelimit: false ratelimit_standard: false retry_after: false exhaustion_status: null note: >- A live tools/list POST to https://mcp.b12.io/mcp returned only cache-control, content-type, date, server (nginx/1.28.1), transfer-encoding and connection. No X-RateLimit-*, no RateLimit-*, no Retry-After. An agent has no runtime backpressure signal on this surface. quotas: - scope: per-account kind: ai-credits plan: Free limit: 5 per day, capped at 30 per month source: https://www.b12.io/pricing/ - scope: per-account kind: ai-credits plan: Website limit: 200 per month source: https://www.b12.io/pricing/ - scope: per-account kind: ai-credits plan: Website + Client Engagement limit: 300 per month source: https://www.b12.io/pricing/ - scope: per-account kind: launched-projects plan: Website limit: 20 source: https://www.b12.io/pricing/ - scope: per-account kind: launched-projects plan: Website + Client Engagement limit: 30 source: https://www.b12.io/pricing/ quota_note: >- These quotas bind to a signed-in B12 account. The MCP tool is called anonymously and returns a signup link rather than a provisioned site, so no quota is enforced at the point an agent calls it, and none is reported back. x-evidence: fetched: '2026-08-13' probes: - url: https://mcp.b12.io/mcp method: POST tools/list http_status: 200 headers_returned: [cache-control, content-type, date, server, transfer-encoding, connection] - url: https://www.b12.io/pricing/ http_status: 200