rateLimits:
  - name: Anaconda.org REST API
    description: >
      The Anaconda.org REST API (api.anaconda.org) does not publish formal
      rate limit documentation for unauthenticated read access to conda-forge
      package metadata. In practice, the API is accessible without
      authentication for GET requests and is used by automated tooling and
      CI pipelines at scale. Anaconda may apply unpublished throttling to
      unauthenticated clients that produce abnormally high request volumes.
      Users automating bulk lookups should add reasonable delays between
      requests and use authenticated sessions with an Anaconda token when
      making large numbers of calls.
    url: https://anaconda.org/conda-forge
    header: ""
    limit: unspecified
    unit: ""
    notes: >
      No SLA is provided for the free community-facing API. Anaconda
      Enterprise customers have separate agreements with defined SLAs.
  - name: conda Channel Repodata (CDN)
    description: >
      The conda channel repodata files (repodata.json, repodata.json.zst,
      channeldata.json) served from conda.anaconda.org are delivered through
      Cloudflare CDN with aggressive caching (cache-control: public,
      max-age=1200). No per-client rate limit is documented. Clients are
      expected to cache index files locally and re-fetch only when stale.
      The conda and mamba clients implement this caching behavior
      automatically. Downloading the full repodata.json for linux-64 is
      ~400 MB uncompressed; the .zst variant is ~50 MB.
    url: https://conda.anaconda.org/conda-forge/
    header: Cache-Control
    limit: unspecified
    unit: ""
    notes: >
      CDN edge caches repodata for 20 minutes (max-age=1200). Clients
      should respect cache headers and avoid unnecessary re-downloads.