generated: '2026-09-02'
method: generated
source: >-
  openapi/standard-compute-openapi.json (operationIds verified against the verbatim
  spec fetched 2026-09-02) plus the provider's getting-started, fair-use, smart-pacing
  and integration pages.
note: >-
  Two skills only. This API has nine operations, five of which are the same inference
  call in different wire dialects, so there are exactly two marquee flows worth
  packaging: getting an agent pointed at the gateway, and not blowing the flat-rate
  budget once it is. Every operationId referenced below was read from the published
  spec; none was invented.
skills:
  - name: standard-compute-switch-agent-provider
    file: standard-compute-switch-agent-provider.md
    description: Repoint an OpenAI- or Anthropic-compatible agent at the flat-rate gateway and verify it.
    operations:
      - list_models_v1_models_get
      - chat_completions_v1_chat_completions_post
  - name: standard-compute-estimate-before-spending
    file: standard-compute-estimate-before-spending.md
    description: Size a request with token counting before it consumes irreversible monthly compute budget.
    operations:
      - anthropic_count_tokens_v1_messages_count_tokens_post
      - list_models_v1_models_get
      - anthropic_messages_v1_messages_post