{
  "id": "token-budget:claude-code-call-default",
  "_kind": "TokenBudget",
  "_file": "lifecycle/token-budgets/claude-code-default.yaml",
  "_cluster": "lifecycle",
  "attributes": {
    "displayName": "Claude Code default per-call token budget",
    "scope": "call",
    "maxTotalTokens": 200000,
    "contextWindowTokens": 200000,
    "floorOutputTokens": 3000,
    "safetyBufferTokens": 1000,
    "enforcement": "downscale-output",
    "onExceededAction": "reduce-max-tokens",
    "description": "Per-call budget enforced by withRetry's max-tokens-context-overflow branch.\nOn a 400 with \"input length and max_tokens exceed context limit\", compute\navailableContext = contextLimit - inputTokens - 1000 safety buffer, then\nset maxTokensOverride = max(FLOOR_OUTPUT_TOKENS, availableContext,\nthinkingBudgetTokens + 1) and retry. If availableContext < FLOOR_OUTPUT_TOKENS\nthe call fails permanently (logError + throw original).\n"
  },
  "outgoingEdges": [],
  "incomingEdges": [
    {
      "from": "transport-client:claude-code-anthropic-direct",
      "to": "token-budget:claude-code-call-default",
      "kind": "client_uses_token_budget",
      "attributes": {}
    }
  ]
}