@bivelio/savings-layer 0.4.16 → 0.4.17

package/dist/index.d.ts

@@ -197,6 +197,16 @@ declare const CATALOGO: Readonly<Record<string, CostModel>>;
  */
 declare function precioDeLista(model: string): CostModel | undefined;
 
+/**
+ * The version the SDK reports for itself.
+ *
+ * Written by hand and deliberately NOT read from package.json: the published
+ * bundle does not ship package.json, so a `require` here would fail at runtime
+ * on the customer's machine. `test/version-declarada.test.ts` ties this
+ * constant to package.json so the two can never drift.
+ */
+declare const BV_SALA_VERSION = "0.4.17";
+
 /**
  * Environment-agnostic UUID generation.
  * ---------------------------------------------------------------------------
@@ -919,6 +929,20 @@ interface SavingsReport {
          * and adding `shadow` to either would be making the savings up.
          */
         basis: CostBasis;
+        /**
+         * Provider-side context clearing, priced at the FLOOR: the input tokens the
+         * provider itself reported as cleared, valued at this model's cached-input
+         * read rate — never at the input rate. Present only when the clearing is
+         * attributable to this layer's context-management settings (the request
+         * did not already carry them) and the model has a cached-input price.
+         *
+         * A separate lane on purpose: it is not part of `baselineEstimate`,
+         * `actual` or `netSavings`, it is never folded into measured money, and it
+         * is not part of the commission base. It is a lower bound of what those
+         * tokens would have cost to keep reading, not a claim about the net effect
+         * of clearing them.
+         */
+        eliminatedInputFloorCost?: number;
     };
     latencyMs: {
         total: number;
@@ -2843,6 +2867,14 @@ interface UsageEmitterOptions {
     now?: () => number;
     /** Whether this process is a CI/ephemeral runner (excluded from billable seats). */
     ci?: boolean;
+    /**
+     * Which surface owns this process: the local gateway (`bvsala claude` and
+     * friends) or the in-process SDK. Sent with the version in the batch's
+     * `x-bvsala-version` header (`gateway/0.4.16`, `sdk/0.4.16`) so the control
+     * plane can tell a gateway that lacks a lever from an SDK that never had it.
+     * Defaults to `"sdk"`; the gateway sets `"gateway"`.
+     */
+    surface?: "gateway" | "sdk";
     /**
      * Se llama cuando el ingest rechaza la credencial (401/403). Existe para que un
      * despliegue con observabilidad pueda ENTERARSE: el aviso por consola sirve en
@@ -3292,33 +3324,4 @@ interface SavingsLayer {
  */
 declare function createSavingsLayer(config: Omit<SavingsLayerConfig, "tokenizer"> & Partial<Pick<SavingsLayerConfig, "tokenizer">>): SavingsLayer;
 
-/**
- * bivelio-savings-layer (BV-SALA)
- * ---------------------------------------------------------------------------
- * PROPRIETARY AND CONFIDENTIAL. Copyright (c) 2026 BiVelio Inc.
- * All Rights Reserved. Unauthorized copying, distribution, or use of this
- * software, in whole or in part, is strictly prohibited. See LICENSE.
- * ---------------------------------------------------------------------------
- *
- * A provider-neutral LLM execution optimizer. For every request it answers, in
- * order: Can I avoid this call? Can I avoid retrieval? Can I send fewer tools?
- * What context is actually needed? Should I compress it? Which format costs
- * least for this tokenizer? What is the minimal output contract? Is the saving
- * still positive after all infrastructure is counted?
- *
- * It keeps FOUR separate ledgers — avoided_calls, eliminated_tokens,
- * compressed_tokens, provider_cached_tokens — because mixing them produces
- * misleading metrics.
- */
-/**
- * The version the SDK reports for itself.
- *
- * Written by hand and deliberately NOT read from package.json: the published
- * bundle does not ship package.json, so a `require` here would fail at runtime
- * on the customer's machine. `test/version-declarada.test.ts` ties this
- * constant to package.json so the two can never drift.
- */
-
-declare const BV_SALA_VERSION = "0.4.16";
-
 export { type AnthropicOptions, AnthropicProvider, BV_SALA_VERSION, type BivelioHandoff, CATALOGO, COMPACT_ROWS_INSTRUCTION, type CacheOutcome, type CacheTier, type CachedAnswer, type CanonicalRequest, type ChatMessage, type CompactRowsDecodeOptions, type CompactRowsOptions, type CompactVerification, type CompressionEconomics, type CompressionPolicyInput, type CompressionRequest, type CompressionResult, type Compressor, type ConstraintBackend, type ConstraintBackendId, type ConstraintPlan, type ConversationState, type CostBasis, type CostModel, DEFAULT_COST_MODEL, DEFAULT_MAX_STALENESS_SECONDS, type DeferredExecutionReport, type DeterministicResolver, EMBEDDED_PUBLIC_KEY, type EmbeddingFn, type EvidenceDoc, type ExactCache, ExactTokenizer, type ExecutionClass, ExecutionRouter, type FormatChoice, FormatSelector, type FormatSelectorOptions, type HandoffFact, HeuristicEstimator, InMemoryExactCache, InMemorySemanticCache, type JsonSchema, LicenseError, type LicenseGate, type LicenseKeyset, type LicenseStatus, type LicenseTokenPayload, type LocalClassifier, MODELOS_POR_CODIFICACION, MODELS_BY_ENCODING, type MemoryDelta, type MessageRole, MockProvider, type MockProviderOptions, type ModelCascadeMode, type ModelCascadeReport, type ModelCascadeSettings, type OpenAICompatibleOptions, OpenAICompatibleProvider, type OutputContract, PRECIOS_AL_DIA, type Policy, type ProtectedSpan, type Provider, type ProviderRequest, type ProviderResponse, type ProviderUsage, QUALITY_STATUSES, type QualityStatus, type RequestContext, type Reranker, type ResolvedDeps, type ResolvedGate, type ResolvedLicense, ResolverRegistry, type ResolverResult, type RetrievalOptions, type RetrievalOutcome, type RetrievalPlan, type Retriever, type RevocationListBody, type RevocationListVerification, type RevocationPollEvent, RevocationPoller, type RevocationPollerOptions, RevocationSnapshot, type RouteDecision, type SavingsLayer, type SavingsLayerConfig, SavingsLedger, type SavingsReport, type SavingsRequest, type SavingsResult, type SavingsToggles, SchemaRegistry, type SemanticCacheOptions, type SemanticHit, type SemanticRecord, type SemanticVerifier, type SerializationFormat, type Serializer, Singleflight, type TokenEstimator, TokenizerRegistry, type ToolCallRequest, type ToolDescriptor, type ToolRisk, type ToolSelection, ToolSelector, type ToolSelectorOptions, TraceCollector, type TraceEvent, UMBRAL_LOCAL, UsageEmitter, type UsageEmitterOptions, type ValidationError, type ValidationResult, adaptiveK, arithmeticResolver, buildExactKey, buildInflightKey, canonicalize, compactRowsSerializer, compresionFiel, computeNextDelay, cosine, createCompactRowsSerializer, createDefaultResolvers, createHandoff, createMockProvider, createSavingsLayer, csvSerializer, decodeCompactRows, dedupe, defaultTokenizerRegistry, deriveSeatId, deriveStatus, detectCi, embebedorLocal, emptyState, encodeCompactRows, evalArithmetic, evaluatePolicy, executeRetrievalPlan, gateFromKey, heuristicEstimator, isKeyset, isStrictCompatible, jsonCompactSerializer, keyTokensMatch, liveGateFromKey, mergeDelta, modelFamilyOf, multisetEqual, negationCount, noopCompressor, normalizeText, numericTokens, precioDeLista, projectContext, providerNativeBackend, randomId, registerExactTokenizer, registrarTokenizadorExacto, renderState, resolveConstraintBackend, resolveDependencies, resolveLicense, sha256, shouldCompress, stableStringify, uniformColumns, validateAgainstSchema, vectorLocal, verifyCompactSignature, verifyLicenseToken, verifyRevocationList, withProtectedSpans, words };