deepline 0.2.15 → 0.2.17

package/dist/bundling-sources/sdk/src/client.ts

@@ -77,6 +77,7 @@ import type {
   ToolMetadata,
   CustomerDbQueryResult,
   DeeplineAgentModelDescription,
+  InferenceQuote,
 } from './types.js';
 import type { MonitorDefinition } from './monitors.js';
 import type { PlayStagedFileRef } from './plays/local-file-discovery.js';
@@ -1919,6 +1920,22 @@ export class DeeplineClient {
     );
   }
 
+  /**
+   * Quote dynamic AI inference pricing for a concrete payload.
+   *
+   * The result separates a planning estimate from a proven authorization
+   * maximum and contains Deepline credits only.
+   */
+  async quoteInferenceTool(
+    toolId: 'ai_inference' | 'deeplineagent',
+    payload: Record<string, unknown>,
+  ): Promise<InferenceQuote> {
+    return this.http.post<InferenceQuote>(
+      `/api/v2/integrations/${encodeURIComponent(toolId)}/quote`,
+      payload,
+    );
+  }
+
   /**
    * Execute a tool and return the standard execution envelope.
    *

package/dist/bundling-sources/sdk/src/release.ts

@@ -160,7 +160,7 @@ export const SDK_RELEASE = {
   // 0.2.0 makes Dataset Handles uniformly async-only after 0.1.320 briefly
   // exposed storage-dependent synchronous access. This deliberate minor
   // release keeps lazy paging semantics independent of row residency.
-  version: '0.2.15',
+  version: '0.2.17',
   contracts: {
     api: {
       name: 'sdk-http-api',

package/dist/bundling-sources/sdk/src/types.ts

@@ -336,8 +336,51 @@ export interface DeeplineAgentModelDescription {
   };
   caveats: string[];
   sources: string[];
+  inferencePricing: InferenceDynamicPricingCapability;
 }
 
+export interface InferenceDynamicPricingCapability {
+  settlement: 'actual_usage';
+  staticUnitPrice: null;
+  quoteEndpointTemplate: string;
+  supportedQualities: Array<'bounded_max' | 'estimate_only' | 'unavailable'>;
+  payloadFields: string[];
+}
+
+export type InferenceQuoteAssumptions = {
+  modelCallCount: number;
+  planningModelCallCount: number;
+  estimatedInputTokens: number;
+  estimatedOutputTokens: number;
+  maxOutputTokens: number | null;
+  candidateRouteCount: number;
+  pricedRouteCount: number;
+  inputBound: 'proven_request_tokens' | 'catalog_context' | 'not_proven';
+  inputBoundProofId?: string;
+};
+
+export type InferenceQuote = {
+  quality: 'bounded_max' | 'estimate_only' | 'unavailable';
+  estimate:
+    | {
+        quality: 'estimate_only';
+        credits: number;
+        assumptions: InferenceQuoteAssumptions;
+      }
+    | { quality: 'unavailable'; assumptions: InferenceQuoteAssumptions };
+  authorization:
+    | {
+        quality: 'bounded_max';
+        maximumCredits: number;
+        assumptions: InferenceQuoteAssumptions;
+      }
+    | {
+        quality: 'estimate_only' | 'unavailable';
+        assumptions: InferenceQuoteAssumptions;
+      };
+  settlement: 'actual_usage';
+};
+
 /**
  * Query options for ranked tool/provider discovery.
  */

package/dist/bundling-sources/shared_libs/play-runtime/run-failure.ts

@@ -1,3 +1,5 @@
+import { ToolExecutionError } from '../tool-execution-error';
+
 const CLOUDFLARE_DURABLE_OBJECT_RESET_RE =
   /Durable Object.*(?:code (?:was|has been) updated|storage caused object)/;
 const CLOUDFLARE_WORKER_SUBREQUEST_LIMIT_RE =
@@ -280,6 +282,20 @@ function toErrorText(error: unknown): string {
 export function normalizePlayRunFailure(error: unknown): PlayRunFailureDetails {
   const rawCause = toErrorText(error);
   const cause = boundedFailureText(rawCause);
+  if (
+    error instanceof ToolExecutionError &&
+    error.origin === 'deepline' &&
+    error.category === 'billing' &&
+    error.code !== 'BILLING_UNAVAILABLE'
+  ) {
+    return {
+      code: error.code?.trim() || 'BILLING_DENIED',
+      phase: 'billing',
+      message: cause,
+      retryable: false,
+      cause,
+    };
+  }
   if (
     (error &&
       typeof error === 'object' &&

package/dist/cli/index.js

@@ -1044,7 +1044,7 @@ var SDK_RELEASE = {
   // 0.2.0 makes Dataset Handles uniformly async-only after 0.1.320 briefly
   // exposed storage-dependent synchronous access. This deliberate minor
   // release keeps lazy paging semantics independent of row residency.
-  version: "0.2.15",
+  version: "0.2.17",
   contracts: {
     api: {
       name: "sdk-http-api",
@@ -4040,6 +4040,18 @@ var DeeplineClient = class {
       }
     );
   }
+  /**
+   * Quote dynamic AI inference pricing for a concrete payload.
+   *
+   * The result separates a planning estimate from a proven authorization
+   * maximum and contains Deepline credits only.
+   */
+  async quoteInferenceTool(toolId, payload) {
+    return this.http.post(
+      `/api/v2/integrations/${encodeURIComponent(toolId)}/quote`,
+      payload
+    );
+  }
   /**
    * Execute a tool and return the standard execution envelope.
    *
@@ -7787,6 +7799,7 @@ async function handleUsage(options) {
   const params = new URLSearchParams();
   if (options.limit) params.set("recent_limit", options.limit);
   if (options.offset) params.set("recent_offset", options.offset);
+  if (options.cursor) params.set("recent_cursor", options.cursor);
   if (options.runId) params.set("run_id", options.runId);
   const suffix = Array.from(params).length > 0 ? `?${params.toString()}` : "";
   const payload = await http.get(
@@ -7804,7 +7817,10 @@ async function handleUsage(options) {
     `Monthly limit: ${quota.enabled ? quota.monthly_credits_limit ?? "(unknown)" : "off"}`,
     ...recentUsageLines(
       Array.isArray(recent.entries) ? recent.entries : []
-    )
+    ),
+    ...typeof recent.next_cursor === "string" && recent.next_cursor.length > 0 ? [
+      `Next page: deepline billing usage --cursor ${recent.next_cursor}${options.limit ? ` --limit ${options.limit}` : ""}`
+    ] : []
   ];
   printCommandEnvelope(
     {
@@ -8527,14 +8543,17 @@ Examples:
     `
 Notes:
   Read-only. Shows last-30-day Deepline credit usage plus a bounded recent-call
-  page. Use --limit/--offset to paginate the recent-call section.
+  page. Use --limit/--cursor to paginate the recent-call section.
 
 Examples:
   deepline billing usage
-  deepline billing usage --limit 50 --offset 50 --json
+  deepline billing usage --limit 50 --cursor <next_cursor> --json
   deepline billing usage --run-id play/example/run/20260630t120000-000-abcdef --json
 `
-  ).option("--limit <n>", "Recent-call page size").option("--offset <n>", "Recent-call offset").option("--run-id <run_id>", "Show recent usage for one play run").option("--json", "Emit JSON output. Also automatic when stdout is piped").action(handleUsage);
+  ).option("--limit <n>", "Recent-call page size").option("--offset <n>", "Recent-call offset").option(
+    "--cursor <cursor>",
+    "Recent-call cursor from the previous response"
+  ).option("--run-id <run_id>", "Show recent usage for one play run").option("--json", "Emit JSON output. Also automatic when stdout is piped").action(handleUsage);
   billing.command("limit").description("Show configured monthly limit state.").addHelpText(
     "after",
     `
@@ -32998,11 +33017,15 @@ Examples:
   deepline tools describe hunter_email_verifier --examples-only
   deepline tools describe hunter_email_verifier --json
   deepline tools describe deeplineagent --model openai/gpt-5.5 --json
+  deepline tools describe ai_inference --estimate-payload @payload.json --json
   deepline tools execute hunter_email_verifier --input '{"email":"a@b.com"}'
 `
   ).option("--json", "Emit JSON output. Also automatic when stdout is piped").option(
     "--model <model>",
     "For deeplineagent, include AI Gateway model/provider option metadata"
+  ).option(
+    "--estimate-payload <payload>",
+    "Quote an ai_inference or deeplineagent JSON payload (inline or @file)"
   ).option("--pricing-only", "Only print pricing and billing semantics").option("--schema-only", "Only print input schema fields").option(
     "--examples-only",
     "Only print runnable examples and sample payloads"
@@ -33025,7 +33048,8 @@ Examples:
       schemaOnly: Boolean(options.schemaOnly),
       examplesOnly: Boolean(options.examplesOnly),
       gettersOnly: Boolean(options.gettersOnly),
-      model: typeof options.model === "string" ? options.model : void 0
+      model: typeof options.model === "string" ? options.model : void 0,
+      estimatePayload: typeof options.estimatePayload === "string" ? options.estimatePayload : void 0
     });
   });
   addToolMetadataCommand(tools.command("describe <toolId>").alias("get"));
@@ -33099,6 +33123,23 @@ async function getTool(toolId, options = {}) {
     }
     throw error;
   }
+  let inferenceQuote = null;
+  if (options.estimatePayload) {
+    if (toolId !== "deeplineagent" && toolId !== "ai_inference") {
+      console.error(
+        "--estimate-payload is only supported for deeplineagent and ai_inference."
+      );
+      return 2;
+    }
+    const estimatePayload = parseJsonObjectArgument(
+      options.estimatePayload,
+      "--estimate-payload"
+    );
+    if (options.model && estimatePayload.model === void 0) {
+      estimatePayload.model = options.model;
+    }
+    inferenceQuote = await client2.quoteInferenceTool(toolId, estimatePayload);
+  }
   const modelDescription = options.model && (toolId === "deeplineagent" || toolId === "ai_inference") ? await client2.describeModel(options.model) : null;
   if (options.model && !modelDescription) {
     console.error(
@@ -33110,7 +33151,8 @@ async function getTool(toolId, options = {}) {
     process.stdout.write(
       `${JSON.stringify({
         ...toolContractJsonForDescribe(tool, toolId),
-        ...modelDescription ? { modelOptions: modelDescription } : {}
+        ...modelDescription ? { modelOptions: modelDescription } : {},
+        ...inferenceQuote ? { inferenceQuote } : {}
       })}
 `
     );
@@ -33121,7 +33163,8 @@ async function getTool(toolId, options = {}) {
     process.stdout.write(
       `${JSON.stringify({
         ...toolMetadataJsonForDescribe(tool, toolId),
-        ...modelDescription ? { modelOptions: modelDescription } : {}
+        ...modelDescription ? { modelOptions: modelDescription } : {},
+        ...inferenceQuote ? { inferenceQuote } : {}
       })}
 `
     );
@@ -33141,18 +33184,22 @@ async function getTool(toolId, options = {}) {
   }
   if (options.pricingOnly) {
     printToolPricingOnly(tool, toolId);
+    if (inferenceQuote) printInferenceQuoteSummary(inferenceQuote);
     return 0;
   }
   if (options.schemaOnly) {
     printToolSchemaOnly(tool, toolId);
+    if (inferenceQuote) printInferenceQuoteSummary(inferenceQuote);
     return 0;
   }
   if (options.examplesOnly) {
     printToolExamplesOnly(tool, toolId);
+    if (inferenceQuote) printInferenceQuoteSummary(inferenceQuote);
     return 0;
   }
   if (options.gettersOnly) {
     printToolGettersOnly(tool, toolId);
+    if (inferenceQuote) printInferenceQuoteSummary(inferenceQuote);
     return 0;
   }
   if (options.compact) {
@@ -33161,13 +33208,15 @@ async function getTool(toolId, options = {}) {
       process.stdout.write(`
 ${formatModelDescription(modelDescription)}`);
     }
+    if (inferenceQuote) printInferenceQuoteSummary(inferenceQuote);
     return 0;
   }
   if (shouldEmitJson()) {
     process.stdout.write(
       `${JSON.stringify({
         ...toolContractJsonForDescribe(tool, toolId),
-        ...modelDescription ? { modelOptions: modelDescription } : {}
+        ...modelDescription ? { modelOptions: modelDescription } : {},
+        ...inferenceQuote ? { inferenceQuote } : {}
       })}
 `
     );
@@ -33178,8 +33227,26 @@ ${formatModelDescription(modelDescription)}`);
     process.stdout.write(`
 ${formatModelDescription(modelDescription)}`);
   }
+  if (inferenceQuote) printInferenceQuoteSummary(inferenceQuote);
   return 0;
 }
+function printInferenceQuoteSummary(quote) {
+  console.log("");
+  console.log("Inference quote:");
+  if (quote.estimate.quality === "estimate_only") {
+    console.log(`- Planning estimate: ${quote.estimate.credits} credits`);
+  } else {
+    console.log("- Planning estimate: unavailable");
+  }
+  if (quote.authorization.quality === "bounded_max") {
+    console.log(
+      `- Maximum authorization: ${quote.authorization.maximumCredits} credits`
+    );
+  } else {
+    console.log(`- Maximum authorization: ${quote.authorization.quality}`);
+  }
+  console.log("- Settlement: actual usage");
+}
 function toolContractJsonForDescribe(tool, requestedToolId) {
   const toolId = String(tool.toolId || requestedToolId);
   const inputFields = toolInputFieldsForDisplay(

package/dist/cli/index.mjs

@@ -1030,7 +1030,7 @@ var SDK_RELEASE = {
   // 0.2.0 makes Dataset Handles uniformly async-only after 0.1.320 briefly
   // exposed storage-dependent synchronous access. This deliberate minor
   // release keeps lazy paging semantics independent of row residency.
-  version: "0.2.15",
+  version: "0.2.17",
   contracts: {
     api: {
       name: "sdk-http-api",
@@ -4026,6 +4026,18 @@ var DeeplineClient = class {
       }
     );
   }
+  /**
+   * Quote dynamic AI inference pricing for a concrete payload.
+   *
+   * The result separates a planning estimate from a proven authorization
+   * maximum and contains Deepline credits only.
+   */
+  async quoteInferenceTool(toolId, payload) {
+    return this.http.post(
+      `/api/v2/integrations/${encodeURIComponent(toolId)}/quote`,
+      payload
+    );
+  }
   /**
    * Execute a tool and return the standard execution envelope.
    *
@@ -7785,6 +7797,7 @@ async function handleUsage(options) {
   const params = new URLSearchParams();
   if (options.limit) params.set("recent_limit", options.limit);
   if (options.offset) params.set("recent_offset", options.offset);
+  if (options.cursor) params.set("recent_cursor", options.cursor);
   if (options.runId) params.set("run_id", options.runId);
   const suffix = Array.from(params).length > 0 ? `?${params.toString()}` : "";
   const payload = await http.get(
@@ -7802,7 +7815,10 @@ async function handleUsage(options) {
     `Monthly limit: ${quota.enabled ? quota.monthly_credits_limit ?? "(unknown)" : "off"}`,
     ...recentUsageLines(
       Array.isArray(recent.entries) ? recent.entries : []
-    )
+    ),
+    ...typeof recent.next_cursor === "string" && recent.next_cursor.length > 0 ? [
+      `Next page: deepline billing usage --cursor ${recent.next_cursor}${options.limit ? ` --limit ${options.limit}` : ""}`
+    ] : []
   ];
   printCommandEnvelope(
     {
@@ -8525,14 +8541,17 @@ Examples:
     `
 Notes:
   Read-only. Shows last-30-day Deepline credit usage plus a bounded recent-call
-  page. Use --limit/--offset to paginate the recent-call section.
+  page. Use --limit/--cursor to paginate the recent-call section.
 
 Examples:
   deepline billing usage
-  deepline billing usage --limit 50 --offset 50 --json
+  deepline billing usage --limit 50 --cursor <next_cursor> --json
   deepline billing usage --run-id play/example/run/20260630t120000-000-abcdef --json
 `
-  ).option("--limit <n>", "Recent-call page size").option("--offset <n>", "Recent-call offset").option("--run-id <run_id>", "Show recent usage for one play run").option("--json", "Emit JSON output. Also automatic when stdout is piped").action(handleUsage);
+  ).option("--limit <n>", "Recent-call page size").option("--offset <n>", "Recent-call offset").option(
+    "--cursor <cursor>",
+    "Recent-call cursor from the previous response"
+  ).option("--run-id <run_id>", "Show recent usage for one play run").option("--json", "Emit JSON output. Also automatic when stdout is piped").action(handleUsage);
   billing.command("limit").description("Show configured monthly limit state.").addHelpText(
     "after",
     `
@@ -33061,11 +33080,15 @@ Examples:
   deepline tools describe hunter_email_verifier --examples-only
   deepline tools describe hunter_email_verifier --json
   deepline tools describe deeplineagent --model openai/gpt-5.5 --json
+  deepline tools describe ai_inference --estimate-payload @payload.json --json
   deepline tools execute hunter_email_verifier --input '{"email":"a@b.com"}'
 `
   ).option("--json", "Emit JSON output. Also automatic when stdout is piped").option(
     "--model <model>",
     "For deeplineagent, include AI Gateway model/provider option metadata"
+  ).option(
+    "--estimate-payload <payload>",
+    "Quote an ai_inference or deeplineagent JSON payload (inline or @file)"
   ).option("--pricing-only", "Only print pricing and billing semantics").option("--schema-only", "Only print input schema fields").option(
     "--examples-only",
     "Only print runnable examples and sample payloads"
@@ -33088,7 +33111,8 @@ Examples:
       schemaOnly: Boolean(options.schemaOnly),
       examplesOnly: Boolean(options.examplesOnly),
       gettersOnly: Boolean(options.gettersOnly),
-      model: typeof options.model === "string" ? options.model : void 0
+      model: typeof options.model === "string" ? options.model : void 0,
+      estimatePayload: typeof options.estimatePayload === "string" ? options.estimatePayload : void 0
     });
   });
   addToolMetadataCommand(tools.command("describe <toolId>").alias("get"));
@@ -33162,6 +33186,23 @@ async function getTool(toolId, options = {}) {
     }
     throw error;
   }
+  let inferenceQuote = null;
+  if (options.estimatePayload) {
+    if (toolId !== "deeplineagent" && toolId !== "ai_inference") {
+      console.error(
+        "--estimate-payload is only supported for deeplineagent and ai_inference."
+      );
+      return 2;
+    }
+    const estimatePayload = parseJsonObjectArgument(
+      options.estimatePayload,
+      "--estimate-payload"
+    );
+    if (options.model && estimatePayload.model === void 0) {
+      estimatePayload.model = options.model;
+    }
+    inferenceQuote = await client2.quoteInferenceTool(toolId, estimatePayload);
+  }
   const modelDescription = options.model && (toolId === "deeplineagent" || toolId === "ai_inference") ? await client2.describeModel(options.model) : null;
   if (options.model && !modelDescription) {
     console.error(
@@ -33173,7 +33214,8 @@ async function getTool(toolId, options = {}) {
     process.stdout.write(
       `${JSON.stringify({
         ...toolContractJsonForDescribe(tool, toolId),
-        ...modelDescription ? { modelOptions: modelDescription } : {}
+        ...modelDescription ? { modelOptions: modelDescription } : {},
+        ...inferenceQuote ? { inferenceQuote } : {}
       })}
 `
     );
@@ -33184,7 +33226,8 @@ async function getTool(toolId, options = {}) {
     process.stdout.write(
       `${JSON.stringify({
         ...toolMetadataJsonForDescribe(tool, toolId),
-        ...modelDescription ? { modelOptions: modelDescription } : {}
+        ...modelDescription ? { modelOptions: modelDescription } : {},
+        ...inferenceQuote ? { inferenceQuote } : {}
       })}
 `
     );
@@ -33204,18 +33247,22 @@ async function getTool(toolId, options = {}) {
   }
   if (options.pricingOnly) {
     printToolPricingOnly(tool, toolId);
+    if (inferenceQuote) printInferenceQuoteSummary(inferenceQuote);
     return 0;
   }
   if (options.schemaOnly) {
     printToolSchemaOnly(tool, toolId);
+    if (inferenceQuote) printInferenceQuoteSummary(inferenceQuote);
     return 0;
   }
   if (options.examplesOnly) {
     printToolExamplesOnly(tool, toolId);
+    if (inferenceQuote) printInferenceQuoteSummary(inferenceQuote);
     return 0;
   }
   if (options.gettersOnly) {
     printToolGettersOnly(tool, toolId);
+    if (inferenceQuote) printInferenceQuoteSummary(inferenceQuote);
     return 0;
   }
   if (options.compact) {
@@ -33224,13 +33271,15 @@ async function getTool(toolId, options = {}) {
       process.stdout.write(`
 ${formatModelDescription(modelDescription)}`);
     }
+    if (inferenceQuote) printInferenceQuoteSummary(inferenceQuote);
     return 0;
   }
   if (shouldEmitJson()) {
     process.stdout.write(
       `${JSON.stringify({
         ...toolContractJsonForDescribe(tool, toolId),
-        ...modelDescription ? { modelOptions: modelDescription } : {}
+        ...modelDescription ? { modelOptions: modelDescription } : {},
+        ...inferenceQuote ? { inferenceQuote } : {}
       })}
 `
     );
@@ -33241,8 +33290,26 @@ ${formatModelDescription(modelDescription)}`);
     process.stdout.write(`
 ${formatModelDescription(modelDescription)}`);
   }
+  if (inferenceQuote) printInferenceQuoteSummary(inferenceQuote);
   return 0;
 }
+function printInferenceQuoteSummary(quote) {
+  console.log("");
+  console.log("Inference quote:");
+  if (quote.estimate.quality === "estimate_only") {
+    console.log(`- Planning estimate: ${quote.estimate.credits} credits`);
+  } else {
+    console.log("- Planning estimate: unavailable");
+  }
+  if (quote.authorization.quality === "bounded_max") {
+    console.log(
+      `- Maximum authorization: ${quote.authorization.maximumCredits} credits`
+    );
+  } else {
+    console.log(`- Maximum authorization: ${quote.authorization.quality}`);
+  }
+  console.log("- Settlement: actual usage");
+}
 function toolContractJsonForDescribe(tool, requestedToolId) {
   const toolId = String(tool.toolId || requestedToolId);
   const inputFields = toolInputFieldsForDisplay(

package/dist/index.d.mts

@@ -411,7 +411,46 @@ interface DeeplineAgentModelDescription {
     };
     caveats: string[];
     sources: string[];
+    inferencePricing: InferenceDynamicPricingCapability;
 }
+interface InferenceDynamicPricingCapability {
+    settlement: 'actual_usage';
+    staticUnitPrice: null;
+    quoteEndpointTemplate: string;
+    supportedQualities: Array<'bounded_max' | 'estimate_only' | 'unavailable'>;
+    payloadFields: string[];
+}
+type InferenceQuoteAssumptions = {
+    modelCallCount: number;
+    planningModelCallCount: number;
+    estimatedInputTokens: number;
+    estimatedOutputTokens: number;
+    maxOutputTokens: number | null;
+    candidateRouteCount: number;
+    pricedRouteCount: number;
+    inputBound: 'proven_request_tokens' | 'catalog_context' | 'not_proven';
+    inputBoundProofId?: string;
+};
+type InferenceQuote = {
+    quality: 'bounded_max' | 'estimate_only' | 'unavailable';
+    estimate: {
+        quality: 'estimate_only';
+        credits: number;
+        assumptions: InferenceQuoteAssumptions;
+    } | {
+        quality: 'unavailable';
+        assumptions: InferenceQuoteAssumptions;
+    };
+    authorization: {
+        quality: 'bounded_max';
+        maximumCredits: number;
+        assumptions: InferenceQuoteAssumptions;
+    } | {
+        quality: 'estimate_only' | 'unavailable';
+        assumptions: InferenceQuoteAssumptions;
+    };
+    settlement: 'actual_usage';
+};
 /**
  * Query options for ranked tool/provider discovery.
  */
@@ -2446,6 +2485,13 @@ declare class DeeplineClient {
      * @returns Model metadata, provider option shapes, and runnable examples
      */
     describeModel(model: string): Promise<DeeplineAgentModelDescription>;
+    /**
+     * Quote dynamic AI inference pricing for a concrete payload.
+     *
+     * The result separates a planning estimate from a proven authorization
+     * maximum and contains Deepline credits only.
+     */
+    quoteInferenceTool(toolId: 'ai_inference' | 'deeplineagent', payload: Record<string, unknown>): Promise<InferenceQuote>;
     /**
      * Execute a tool and return the standard execution envelope.
      *

package/dist/index.d.ts

@@ -411,7 +411,46 @@ interface DeeplineAgentModelDescription {
     };
     caveats: string[];
     sources: string[];
+    inferencePricing: InferenceDynamicPricingCapability;
 }
+interface InferenceDynamicPricingCapability {
+    settlement: 'actual_usage';
+    staticUnitPrice: null;
+    quoteEndpointTemplate: string;
+    supportedQualities: Array<'bounded_max' | 'estimate_only' | 'unavailable'>;
+    payloadFields: string[];
+}
+type InferenceQuoteAssumptions = {
+    modelCallCount: number;
+    planningModelCallCount: number;
+    estimatedInputTokens: number;
+    estimatedOutputTokens: number;
+    maxOutputTokens: number | null;
+    candidateRouteCount: number;
+    pricedRouteCount: number;
+    inputBound: 'proven_request_tokens' | 'catalog_context' | 'not_proven';
+    inputBoundProofId?: string;
+};
+type InferenceQuote = {
+    quality: 'bounded_max' | 'estimate_only' | 'unavailable';
+    estimate: {
+        quality: 'estimate_only';
+        credits: number;
+        assumptions: InferenceQuoteAssumptions;
+    } | {
+        quality: 'unavailable';
+        assumptions: InferenceQuoteAssumptions;
+    };
+    authorization: {
+        quality: 'bounded_max';
+        maximumCredits: number;
+        assumptions: InferenceQuoteAssumptions;
+    } | {
+        quality: 'estimate_only' | 'unavailable';
+        assumptions: InferenceQuoteAssumptions;
+    };
+    settlement: 'actual_usage';
+};
 /**
  * Query options for ranked tool/provider discovery.
  */
@@ -2446,6 +2485,13 @@ declare class DeeplineClient {
      * @returns Model metadata, provider option shapes, and runnable examples
      */
     describeModel(model: string): Promise<DeeplineAgentModelDescription>;
+    /**
+     * Quote dynamic AI inference pricing for a concrete payload.
+     *
+     * The result separates a planning estimate from a proven authorization
+     * maximum and contains Deepline credits only.
+     */
+    quoteInferenceTool(toolId: 'ai_inference' | 'deeplineagent', payload: Record<string, unknown>): Promise<InferenceQuote>;
     /**
      * Execute a tool and return the standard execution envelope.
      *

package/dist/index.js

@@ -763,7 +763,7 @@ var SDK_RELEASE = {
   // 0.2.0 makes Dataset Handles uniformly async-only after 0.1.320 briefly
   // exposed storage-dependent synchronous access. This deliberate minor
   // release keeps lazy paging semantics independent of row residency.
-  version: "0.2.15",
+  version: "0.2.17",
   contracts: {
     api: {
       name: "sdk-http-api",
@@ -3759,6 +3759,18 @@ var DeeplineClient = class {
       }
     );
   }
+  /**
+   * Quote dynamic AI inference pricing for a concrete payload.
+   *
+   * The result separates a planning estimate from a proven authorization
+   * maximum and contains Deepline credits only.
+   */
+  async quoteInferenceTool(toolId, payload) {
+    return this.http.post(
+      `/api/v2/integrations/${encodeURIComponent(toolId)}/quote`,
+      payload
+    );
+  }
   /**
    * Execute a tool and return the standard execution envelope.
    *

package/dist/index.mjs

@@ -689,7 +689,7 @@ var SDK_RELEASE = {
   // 0.2.0 makes Dataset Handles uniformly async-only after 0.1.320 briefly
   // exposed storage-dependent synchronous access. This deliberate minor
   // release keeps lazy paging semantics independent of row residency.
-  version: "0.2.15",
+  version: "0.2.17",
   contracts: {
     api: {
       name: "sdk-http-api",
@@ -3685,6 +3685,18 @@ var DeeplineClient = class {
       }
     );
   }
+  /**
+   * Quote dynamic AI inference pricing for a concrete payload.
+   *
+   * The result separates a planning estimate from a proven authorization
+   * maximum and contains Deepline credits only.
+   */
+  async quoteInferenceTool(toolId, payload) {
+    return this.http.post(
+      `/api/v2/integrations/${encodeURIComponent(toolId)}/quote`,
+      payload
+    );
+  }
   /**
    * Execute a tool and return the standard execution envelope.
    *

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "deepline",
-  "version": "0.2.15",
+  "version": "0.2.17",
   "description": "Deepline SDK + CLI — B2B data enrichment powered by durable cloud execution",
   "license": "MIT",
   "repository": {