# Provenance stamped 2026-10-07 (roadmap#943): written by an API Evangelist sweep (dated 2026-05-23 by the file itself), not # harvested from the provider. `method: generated` so the economics campaign may replace it with a # dated, verbatim transcription; every figure below is unverified until it does. method: generated generated: '2026-05-23' specification: API Commons Rate Limits specificationVersion: '0.1' schema: https://raw.githubusercontent.com/api-evangelist/interface-research/main/schema/api-commons.yml#/$defs/RateLimits provider: McKinsey & Company providerId: mckinsey-and-company created: '2026-05-23' modified: '2026-05-23' reconciled: false tags: - Consulting - Management Consulting - Professional Services - MBB - Rate Limiting description: >- McKinsey & Company does not publish a public developer API, so no public rate limits exist. Throughput, concurrency, and SLAs on any McKinsey-built API (within a client engagement, a McKinsey Digital accelerator deployment, a QuantumBlack analytics product, or a managed-services arrangement) are negotiated bilaterally and owned by the engagement contract. notes: No public developer API — placeholder only. sources: - https://www.mckinsey.com/ - https://github.com/mckinsey responseCodes: throttled: 429 limits: - name: No public rate limits scope: account metric: not_applicable limit: not applicable policies: - name: Engagement-defined description: >- Throughput, concurrency, and SLAs on any McKinsey-built API are defined inside the client engagement contract or the relevant managed-services agreement, not by a public developer-portal policy. - name: Open-source self-hosted description: >- Rate limits on McKinsey's open-source projects (Vizro, Agents-at-Scale ARK, ARK Marketplace, CausalNex, QuantumBlack Design System, qbstyles, Kedro) are determined entirely by the adopter's own infrastructure — there is no hosted McKinsey offering. maintainers: - FN: Kin Lane email: kin@apievangelist.com