@sohqureshi/tokenwise 1.0.7 → 1.0.10

package/dist/core/analyze.cjs

@@ -150,8 +150,18 @@ function serializeForTokenEstimate(value) {
 function estimateTokensHeuristic(value) {
   return Math.ceil(serializeForTokenEstimate(value).length / 4);
 }
+function expectedEncodingForModel(model) {
+  const normalizedModel = model.toLowerCase();
+  if (normalizedModel.includes("gpt-4o") || normalizedModel.includes("gpt-4.1") || normalizedModel.includes("o1") || normalizedModel.includes("o3") || normalizedModel.includes("o4")) {
+    return "o200k_base";
+  }
+  if (normalizedModel.includes("davinci") || normalizedModel.startsWith("text-") || normalizedModel.includes("babbage") || normalizedModel.includes("curie")) {
+    return "r50k_base";
+  }
+  return "cl100k_base";
+}
 function estimateTokensWithMeta(value, options = {}) {
-  const { exact = false, model = "gpt-4o-mini", fallbackToHeuristic = true } = options;
+  const { exact = true, model = "gpt-4o-mini", fallbackToHeuristic = true } = options;
   const heuristic = estimateTokensHeuristic(value);
   const heuristicEstimator = "heuristic: 1 token \u2248 4 characters";
   if (!exact) {
@@ -160,36 +170,40 @@ function estimateTokensWithMeta(value, options = {}) {
   try {
     const tiktoken = require2("tiktoken");
     const encoder = tiktoken.encoding_for_model(model);
-    const text = serializeForTokenEstimate(value);
-    const count = encoder.encode(text).length;
-    let encodingName = null;
-    if (encoder.name) encodingName = encoder.name;
+    const count = encoder.encode(serializeForTokenEstimate(value)).length;
+    let encodingName = encoder.name;
     if (!encodingName && typeof tiktoken.model_to_encoding === "function") {
       try {
         encodingName = tiktoken.model_to_encoding(model);
-      } catch (e) {
-        encodingName = null;
+      } catch {
+        encodingName = void 0;
       }
     }
-    if (!encodingName) {
-      const m = String(model || "").toLowerCase();
-      if (m.includes("davinci") || m.startsWith("text-")) encodingName = "r50k_base";
-      else encodingName = "cl100k_base";
-    }
-    const estimator = `exact tokenizer: model=${model} encoding=${encodingName}`;
-    return { count, estimator };
-  } catch (e) {
+    encodingName ?? (encodingName = expectedEncodingForModel(model));
+    encoder.free?.();
+    return {
+      count,
+      estimator: `exact tokenizer: model=${model} encoding=${encodingName}`
+    };
+  } catch {
+    const expectedEncoding = expectedEncodingForModel(model);
     if (!fallbackToHeuristic) {
-      return { count: heuristic, estimator: "exact requested but tokenizer unavailable" };
+      return {
+        count: heuristic,
+        estimator: `exact requested but tokenizer unavailable (expected encoding=${expectedEncoding} for model=${model})`
+      };
     }
-    return { count: heuristic, estimator: heuristicEstimator };
+    return {
+      count: heuristic,
+      estimator: `${heuristicEstimator} (model=${model} expected_encoding=${expectedEncoding})`
+    };
   }
 }
 
 // src/core/analyze.ts
 function analyze(input, options = {}) {
   const {
-    exact = false,
+    exact = true,
     model = "gpt-4o-mini",
     fallbackToHeuristic = true
   } = options;

package/dist/core/analyze.js

@@ -1,11 +1,11 @@
 import {
   analyze
-} from "../chunk-IOMOVRC7.js";
+} from "../chunk-2TDVL6BJ.js";
 import "../chunk-6VRLDRJK.js";
 import "../chunk-XE36GLJP.js";
 import "../chunk-L7BC62MT.js";
 import "../chunk-ZD536GZF.js";
-import "../chunk-U433WUUT.js";
+import "../chunk-L3HY6AWL.js";
 export {
   analyze
 };

package/dist/core/natural.cjs

@@ -42,6 +42,15 @@ function toNatural(data, depth = 0) {
   return String(data);
 }
 function buildContextualStory(obj, depth = 0) {
+  const semanticStory = buildSemanticStory(obj);
+  if (semanticStory) return semanticStory;
+  const nestedEntity = Object.values(obj).find((value) => {
+    return isPlainObject(value) && buildSemanticStory(value) !== null;
+  });
+  if (nestedEntity) {
+    const nestedStory = buildSemanticStory(nestedEntity);
+    if (nestedStory) return nestedStory;
+  }
   let name = obj.name || obj.userName || obj.user;
   if (typeof name === "object" && name !== null && name.name) {
     name = name.name;
@@ -66,6 +75,133 @@ function buildContextualStory(obj, depth = 0) {
   }
   return story + ".";
 }
+function buildSemanticStory(obj) {
+  if (typeof obj.holderName === "string" && typeof obj.policyNumber === "string") {
+    const subject2 = `${obj.holderName} has policy ${obj.policyNumber}`;
+    const details = [
+      typeof obj.planType === "string" ? `a ${obj.planType} plan` : null,
+      typeof obj.premiumAmount === "number" ? `with a premium of ${formatNumber(obj.premiumAmount)}` : null
+    ].filter((value) => value !== null);
+    const sentences = [`${subject2}${details.length ? `, ${details.join(", ")}` : ""}.`];
+    if (isPlainObject(obj.claim)) {
+      const claim = describeClaim(obj.claim);
+      if (claim) sentences.push(claim);
+    }
+    if (Array.isArray(obj.dependents) && obj.dependents.length > 0) {
+      sentences.push(`${obj.holderName}'s dependents are ${joinNaturalList(obj.dependents.map(String))}.`);
+    }
+    return sentences.join(" ");
+  }
+  const subject = findSubject(obj);
+  const stateEntry = findSemanticEntry(obj, [
+    "status",
+    "state",
+    "condition",
+    "stage",
+    "role",
+    "type",
+    "category",
+    "classification",
+    "priority",
+    "phase",
+    "mode",
+    "availability",
+    "outcome",
+    "result",
+    "health",
+    "progress",
+    "visibility",
+    "access",
+    "membership",
+    "sentiment",
+    "severity"
+  ]);
+  if (subject && stateEntry && typeof stateEntry[1] === "string") {
+    const details = Object.entries(obj).filter(([key, value]) => key !== stateEntry[0] && key !== subject.key && shouldDescribeSemantically(key, value)).map(([key, value]) => `${camelToWords(key)} ${formatSemanticValue(value)}`);
+    return `${subject.value} is ${stateEntry[1]}${details.length ? ` and has ${details.join(", ")}` : ""}.`;
+  }
+  return null;
+}
+function findSubject(obj) {
+  const subjectKeys = [
+    "name",
+    "title",
+    "label",
+    "displayName",
+    "entityName",
+    "fullName",
+    "userName",
+    "username",
+    "personName",
+    "customerName",
+    "clientName",
+    "ownerName",
+    "accountName",
+    "companyName",
+    "organizationName",
+    "teamName",
+    "departmentName",
+    "projectName",
+    "productName",
+    "serviceName",
+    "resourceName",
+    "fileName",
+    "deviceName",
+    "hostName",
+    "applicationName",
+    "appName",
+    "taskName",
+    "eventName",
+    "itemName",
+    "orderName",
+    "patientName",
+    "employeeName"
+  ];
+  const entries = Object.entries(obj);
+  for (const key of subjectKeys) {
+    if (typeof obj[key] === "string" && obj[key].trim()) {
+      return { key, value: obj[key].trim() };
+    }
+  }
+  const nameEntry = entries.find(([key, value]) => {
+    return key.toLowerCase().includes("name") && typeof value === "string" && value.trim().length > 0;
+  });
+  if (nameEntry) {
+    return { key: nameEntry[0], value: nameEntry[1].trim() };
+  }
+  return null;
+}
+function findSemanticEntry(obj, keys) {
+  const entries = Object.entries(obj);
+  const exactEntry = entries.find(([key]) => keys.includes(key));
+  if (exactEntry) return exactEntry;
+  return entries.find(([key]) => {
+    const normalizedKey = key.toLowerCase();
+    return keys.some((semanticKey) => normalizedKey.includes(semanticKey.toLowerCase()));
+  }) ?? null;
+}
+function shouldDescribeSemantically(key, value) {
+  return (typeof value === "string" || typeof value === "number" || typeof value === "boolean") && !["id", "timestamp", "createdAt", "updatedAt", "apiKey", "debug"].includes(key);
+}
+function formatSemanticValue(value) {
+  if (typeof value === "number") return formatNumber(value);
+  if (typeof value === "boolean") return value ? "enabled" : "disabled";
+  return String(value);
+}
+function describeClaim(claim) {
+  const parts = [];
+  if (typeof claim.status === "string") parts.push(`The claim is ${claim.status}`);
+  if (typeof claim.requestedAmount === "number") {
+    parts.push(`for ${formatNumber(claim.requestedAmount)}`);
+  }
+  if (typeof claim.claimNumber === "string") {
+    parts.push(`(reference ${claim.claimNumber})`);
+  }
+  return parts.length ? `${parts.join(" ")}.` : null;
+}
+function formatNumber(value) {
+  return new Intl.NumberFormat("en-US").format(value);
+}
 function formatPropertyClause(key, value, depth) {
   const naturalKey = camelToWords(key);
   if (typeof value === "object" && value !== null && !Array.isArray(value)) {

package/dist/core/natural.js

@@ -1,6 +1,6 @@
 import {
   toNatural
-} from "../chunk-D4CFTFM3.js";
+} from "../chunk-BN3L7TUV.js";
 export {
   toNatural
 };

package/dist/core/token.cjs

@@ -36,11 +36,21 @@ function serializeForTokenEstimate(value) {
 function estimateTokensHeuristic(value) {
   return Math.ceil(serializeForTokenEstimate(value).length / 4);
 }
+function expectedEncodingForModel(model) {
+  const normalizedModel = model.toLowerCase();
+  if (normalizedModel.includes("gpt-4o") || normalizedModel.includes("gpt-4.1") || normalizedModel.includes("o1") || normalizedModel.includes("o3") || normalizedModel.includes("o4")) {
+    return "o200k_base";
+  }
+  if (normalizedModel.includes("davinci") || normalizedModel.startsWith("text-") || normalizedModel.includes("babbage") || normalizedModel.includes("curie")) {
+    return "r50k_base";
+  }
+  return "cl100k_base";
+}
 function estimateTokens(value, options = {}) {
   return estimateTokensWithMeta(value, options).count;
 }
 function estimateTokensWithMeta(value, options = {}) {
-  const { exact = false, model = "gpt-4o-mini", fallbackToHeuristic = true } = options;
+  const { exact = true, model = "gpt-4o-mini", fallbackToHeuristic = true } = options;
   const heuristic = estimateTokensHeuristic(value);
   const heuristicEstimator = "heuristic: 1 token \u2248 4 characters";
   if (!exact) {
@@ -49,29 +59,33 @@ function estimateTokensWithMeta(value, options = {}) {
   try {
     const tiktoken = require2("tiktoken");
     const encoder = tiktoken.encoding_for_model(model);
-    const text = serializeForTokenEstimate(value);
-    const count = encoder.encode(text).length;
-    let encodingName = null;
-    if (encoder.name) encodingName = encoder.name;
+    const count = encoder.encode(serializeForTokenEstimate(value)).length;
+    let encodingName = encoder.name;
     if (!encodingName && typeof tiktoken.model_to_encoding === "function") {
       try {
         encodingName = tiktoken.model_to_encoding(model);
-      } catch (e) {
-        encodingName = null;
+      } catch {
+        encodingName = void 0;
       }
     }
-    if (!encodingName) {
-      const m = String(model || "").toLowerCase();
-      if (m.includes("davinci") || m.startsWith("text-")) encodingName = "r50k_base";
-      else encodingName = "cl100k_base";
-    }
-    const estimator = `exact tokenizer: model=${model} encoding=${encodingName}`;
-    return { count, estimator };
-  } catch (e) {
+    encodingName ?? (encodingName = expectedEncodingForModel(model));
+    encoder.free?.();
+    return {
+      count,
+      estimator: `exact tokenizer: model=${model} encoding=${encodingName}`
+    };
+  } catch {
+    const expectedEncoding = expectedEncodingForModel(model);
     if (!fallbackToHeuristic) {
-      return { count: heuristic, estimator: "exact requested but tokenizer unavailable" };
+      return {
+        count: heuristic,
+        estimator: `exact requested but tokenizer unavailable (expected encoding=${expectedEncoding} for model=${model})`
+      };
     }
-    return { count: heuristic, estimator: heuristicEstimator };
+    return {
+      count: heuristic,
+      estimator: `${heuristicEstimator} (model=${model} expected_encoding=${expectedEncoding})`
+    };
   }
 }
 // Annotate the CommonJS export names for ESM import in node:

package/dist/core/token.d.cts

@@ -4,20 +4,14 @@ interface TokenEstimateOptions {
     fallbackToHeuristic?: boolean;
 }
 declare function serializeForTokenEstimate(value: unknown): string;
-/**
- * Estimates tokens using a transparent four-characters-per-token heuristic.
- * JSON values are serialized first, matching the compact form normally sent
- * to an API. Set exact=true to attempt a model-specific tokenizer count when
- * the runtime has the tokenizer installed.
- */
 type TokenEstimateResult = {
     count: number;
     estimator: string;
 };
 declare function estimateTokens(value: unknown, options?: TokenEstimateOptions): number;
 /**
- * Returns both a token count and an estimator string describing how the count was
- * obtained (exact tokenizer + encoding when available, or heuristic otherwise).
+ * Returns an exact model tokenizer count by default. Set exact=false to use the
+ * lightweight four-characters-per-token fallback explicitly.
  */
 declare function estimateTokensWithMeta(value: unknown, options?: TokenEstimateOptions): TokenEstimateResult;
 

package/dist/core/token.d.ts

@@ -4,20 +4,14 @@ interface TokenEstimateOptions {
     fallbackToHeuristic?: boolean;
 }
 declare function serializeForTokenEstimate(value: unknown): string;
-/**
- * Estimates tokens using a transparent four-characters-per-token heuristic.
- * JSON values are serialized first, matching the compact form normally sent
- * to an API. Set exact=true to attempt a model-specific tokenizer count when
- * the runtime has the tokenizer installed.
- */
 type TokenEstimateResult = {
     count: number;
     estimator: string;
 };
 declare function estimateTokens(value: unknown, options?: TokenEstimateOptions): number;
 /**
- * Returns both a token count and an estimator string describing how the count was
- * obtained (exact tokenizer + encoding when available, or heuristic otherwise).
+ * Returns an exact model tokenizer count by default. Set exact=false to use the
+ * lightweight four-characters-per-token fallback explicitly.
  */
 declare function estimateTokensWithMeta(value: unknown, options?: TokenEstimateOptions): TokenEstimateResult;
 

package/dist/core/token.js

@@ -2,7 +2,7 @@ import {
   estimateTokens,
   estimateTokensWithMeta,
   serializeForTokenEstimate
-} from "../chunk-U433WUUT.js";
+} from "../chunk-L3HY6AWL.js";
 export {
   estimateTokens,
   estimateTokensWithMeta,

package/dist/index.cjs

@@ -159,11 +159,21 @@ function serializeForTokenEstimate(value) {
 function estimateTokensHeuristic(value) {
   return Math.ceil(serializeForTokenEstimate(value).length / 4);
 }
+function expectedEncodingForModel(model) {
+  const normalizedModel = model.toLowerCase();
+  if (normalizedModel.includes("gpt-4o") || normalizedModel.includes("gpt-4.1") || normalizedModel.includes("o1") || normalizedModel.includes("o3") || normalizedModel.includes("o4")) {
+    return "o200k_base";
+  }
+  if (normalizedModel.includes("davinci") || normalizedModel.startsWith("text-") || normalizedModel.includes("babbage") || normalizedModel.includes("curie")) {
+    return "r50k_base";
+  }
+  return "cl100k_base";
+}
 function estimateTokens(value, options = {}) {
   return estimateTokensWithMeta(value, options).count;
 }
 function estimateTokensWithMeta(value, options = {}) {
-  const { exact = false, model = "gpt-4o-mini", fallbackToHeuristic = true } = options;
+  const { exact = true, model = "gpt-4o-mini", fallbackToHeuristic = true } = options;
   const heuristic = estimateTokensHeuristic(value);
   const heuristicEstimator = "heuristic: 1 token \u2248 4 characters";
   if (!exact) {
@@ -172,36 +182,40 @@ function estimateTokensWithMeta(value, options = {}) {
   try {
     const tiktoken = require2("tiktoken");
     const encoder = tiktoken.encoding_for_model(model);
-    const text = serializeForTokenEstimate(value);
-    const count = encoder.encode(text).length;
-    let encodingName = null;
-    if (encoder.name) encodingName = encoder.name;
+    const count = encoder.encode(serializeForTokenEstimate(value)).length;
+    let encodingName = encoder.name;
     if (!encodingName && typeof tiktoken.model_to_encoding === "function") {
       try {
         encodingName = tiktoken.model_to_encoding(model);
-      } catch (e) {
-        encodingName = null;
+      } catch {
+        encodingName = void 0;
       }
     }
-    if (!encodingName) {
-      const m = String(model || "").toLowerCase();
-      if (m.includes("davinci") || m.startsWith("text-")) encodingName = "r50k_base";
-      else encodingName = "cl100k_base";
-    }
-    const estimator = `exact tokenizer: model=${model} encoding=${encodingName}`;
-    return { count, estimator };
-  } catch (e) {
+    encodingName ?? (encodingName = expectedEncodingForModel(model));
+    encoder.free?.();
+    return {
+      count,
+      estimator: `exact tokenizer: model=${model} encoding=${encodingName}`
+    };
+  } catch {
+    const expectedEncoding = expectedEncodingForModel(model);
     if (!fallbackToHeuristic) {
-      return { count: heuristic, estimator: "exact requested but tokenizer unavailable" };
+      return {
+        count: heuristic,
+        estimator: `exact requested but tokenizer unavailable (expected encoding=${expectedEncoding} for model=${model})`
+      };
     }
-    return { count: heuristic, estimator: heuristicEstimator };
+    return {
+      count: heuristic,
+      estimator: `${heuristicEstimator} (model=${model} expected_encoding=${expectedEncoding})`
+    };
   }
 }
 
 // src/core/analyze.ts
 function analyze(input, options = {}) {
   const {
-    exact = false,
+    exact = true,
     model = "gpt-4o-mini",
     fallbackToHeuristic = true
   } = options;
@@ -280,6 +294,15 @@ function toNatural(data, depth = 0) {
   return String(data);
 }
 function buildContextualStory(obj, depth = 0) {
+  const semanticStory = buildSemanticStory(obj);
+  if (semanticStory) return semanticStory;
+  const nestedEntity = Object.values(obj).find((value) => {
+    return isPlainObject2(value) && buildSemanticStory(value) !== null;
+  });
+  if (nestedEntity) {
+    const nestedStory = buildSemanticStory(nestedEntity);
+    if (nestedStory) return nestedStory;
+  }
   let name = obj.name || obj.userName || obj.user;
   if (typeof name === "object" && name !== null && name.name) {
     name = name.name;
@@ -304,6 +327,133 @@ function buildContextualStory(obj, depth = 0) {
   }
   return story + ".";
 }
+function buildSemanticStory(obj) {
+  if (typeof obj.holderName === "string" && typeof obj.policyNumber === "string") {
+    const subject2 = `${obj.holderName} has policy ${obj.policyNumber}`;
+    const details = [
+      typeof obj.planType === "string" ? `a ${obj.planType} plan` : null,
+      typeof obj.premiumAmount === "number" ? `with a premium of ${formatNumber(obj.premiumAmount)}` : null
+    ].filter((value) => value !== null);
+    const sentences = [`${subject2}${details.length ? `, ${details.join(", ")}` : ""}.`];
+    if (isPlainObject2(obj.claim)) {
+      const claim = describeClaim(obj.claim);
+      if (claim) sentences.push(claim);
+    }
+    if (Array.isArray(obj.dependents) && obj.dependents.length > 0) {
+      sentences.push(`${obj.holderName}'s dependents are ${joinNaturalList(obj.dependents.map(String))}.`);
+    }
+    return sentences.join(" ");
+  }
+  const subject = findSubject(obj);
+  const stateEntry = findSemanticEntry(obj, [
+    "status",
+    "state",
+    "condition",
+    "stage",
+    "role",
+    "type",
+    "category",
+    "classification",
+    "priority",
+    "phase",
+    "mode",
+    "availability",
+    "outcome",
+    "result",
+    "health",
+    "progress",
+    "visibility",
+    "access",
+    "membership",
+    "sentiment",
+    "severity"
+  ]);
+  if (subject && stateEntry && typeof stateEntry[1] === "string") {
+    const details = Object.entries(obj).filter(([key, value]) => key !== stateEntry[0] && key !== subject.key && shouldDescribeSemantically(key, value)).map(([key, value]) => `${camelToWords(key)} ${formatSemanticValue(value)}`);
+    return `${subject.value} is ${stateEntry[1]}${details.length ? ` and has ${details.join(", ")}` : ""}.`;
+  }
+  return null;
+}
+function findSubject(obj) {
+  const subjectKeys = [
+    "name",
+    "title",
+    "label",
+    "displayName",
+    "entityName",
+    "fullName",
+    "userName",
+    "username",
+    "personName",
+    "customerName",
+    "clientName",
+    "ownerName",
+    "accountName",
+    "companyName",
+    "organizationName",
+    "teamName",
+    "departmentName",
+    "projectName",
+    "productName",
+    "serviceName",
+    "resourceName",
+    "fileName",
+    "deviceName",
+    "hostName",
+    "applicationName",
+    "appName",
+    "taskName",
+    "eventName",
+    "itemName",
+    "orderName",
+    "patientName",
+    "employeeName"
+  ];
+  const entries = Object.entries(obj);
+  for (const key of subjectKeys) {
+    if (typeof obj[key] === "string" && obj[key].trim()) {
+      return { key, value: obj[key].trim() };
+    }
+  }
+  const nameEntry = entries.find(([key, value]) => {
+    return key.toLowerCase().includes("name") && typeof value === "string" && value.trim().length > 0;
+  });
+  if (nameEntry) {
+    return { key: nameEntry[0], value: nameEntry[1].trim() };
+  }
+  return null;
+}
+function findSemanticEntry(obj, keys) {
+  const entries = Object.entries(obj);
+  const exactEntry = entries.find(([key]) => keys.includes(key));
+  if (exactEntry) return exactEntry;
+  return entries.find(([key]) => {
+    const normalizedKey = key.toLowerCase();
+    return keys.some((semanticKey) => normalizedKey.includes(semanticKey.toLowerCase()));
+  }) ?? null;
+}
+function shouldDescribeSemantically(key, value) {
+  return (typeof value === "string" || typeof value === "number" || typeof value === "boolean") && !["id", "timestamp", "createdAt", "updatedAt", "apiKey", "debug"].includes(key);
+}
+function formatSemanticValue(value) {
+  if (typeof value === "number") return formatNumber(value);
+  if (typeof value === "boolean") return value ? "enabled" : "disabled";
+  return String(value);
+}
+function describeClaim(claim) {
+  const parts = [];
+  if (typeof claim.status === "string") parts.push(`The claim is ${claim.status}`);
+  if (typeof claim.requestedAmount === "number") {
+    parts.push(`for ${formatNumber(claim.requestedAmount)}`);
+  }
+  if (typeof claim.claimNumber === "string") {
+    parts.push(`(reference ${claim.claimNumber})`);
+  }
+  return parts.length ? `${parts.join(" ")}.` : null;
+}
+function formatNumber(value) {
+  return new Intl.NumberFormat("en-US").format(value);
+}
 function formatPropertyClause(key, value, depth) {
   const naturalKey = camelToWords(key);
   if (typeof value === "object" && value !== null && !Array.isArray(value)) {

package/dist/index.js

@@ -1,9 +1,9 @@
 import {
   AIChain
-} from "./chunk-P6KXISNT.js";
+} from "./chunk-FBJ3DHS6.js";
 import {
   analyze
-} from "./chunk-IOMOVRC7.js";
+} from "./chunk-2TDVL6BJ.js";
 import {
   toTOON
 } from "./chunk-6VRLDRJK.js";
@@ -15,14 +15,14 @@ import {
 } from "./chunk-L7BC62MT.js";
 import {
   toNatural
-} from "./chunk-D4CFTFM3.js";
+} from "./chunk-BN3L7TUV.js";
 import {
   prune
 } from "./chunk-ZD536GZF.js";
 import {
   estimateTokens,
   serializeForTokenEstimate
-} from "./chunk-U433WUUT.js";
+} from "./chunk-L3HY6AWL.js";
 
 // src/index.ts
 function ai(data) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sohqureshi/tokenwise",
-  "version": "1.0.7",
+  "version": "1.0.10",
   "description": "Optimize JSON data for AI by reducing token usage",
   "repository": {
     "type": "git",