@absolutejs/ai 0.1.1 → 0.1.3

package/CHANGELOG.md

@@ -6,6 +6,18 @@ This file is generated by `absolute-changelog` from the entries in
 `changelog/`. Edit an entry, not this file — and add new ones under
 `changelog/unreleased/`.
 
+## 0.1.3 — 2026-09-16
+
+### Fixed
+
+- **Exclude OpenAI cache-write tokens from ordinary input so cache-aware billing never charges the same tokens twice** (`openai`, `openaiResponses`)
+
+## 0.1.2 — 2026-09-16
+
+### Fixed
+
+- **Preserve every parallel OpenAI tool result and subsequent conversation turn without overwriting expanded messages** (`openai`)
+
 ## 0.1.1 — 2026-09-16
 
 ### Added

package/changelog.json

@@ -1,121 +1,148 @@
 {
-  "contract": 1,
-  "name": "@absolutejs/ai",
-  "releases": [
-    {
-      "changes": [
-        {
-          "kind": "added",
-          "summary": "Allow a separate preparation model with its own verified capacity while preserving the final answer model and invalidating notes when the preparation model changes",
-          "symbols": [
-            "PrepareAITextInputOptions",
-            "prepareAITextInput",
-            "prepareAITextInputStep"
-          ]
-        },
-        {
-          "kind": "fixed",
-          "summary": "Fill preparation sections with bounded tokenizer-checked refinement, avoid duplicate final counting, and honor cancellation before generation",
-          "symbols": [
-            "prepareAITextInput",
-            "prepareAITextInputStep"
-          ]
-        }
-      ],
-      "date": "2026-09-16",
-      "version": "0.1.1"
-    },
-    {
-      "changes": [
-        {
-          "kind": "changed",
-          "summary": "Validate model capacity by default across generation, structured output, WebSocket and SSE entry points; reserve token headroom and support one protocol-preserving context recovery attempt. Providers without counting capability now fail explicitly; contextPolicy: false opts into raw behavior.",
-          "symbols": [
-            "generateAI",
-            "generateAIWithTools",
-            "generateObjectAI",
-            "streamAI",
-            "streamAIToSSE",
-            "streamAIWithTools",
-            "inspectAIContext",
-            "createAIStoredToolResultRecovery",
-            "StreamAIWithToolsOptions",
-            "GenerateAIOptions",
-            "GenerateAIResult",
-            "GenerateObjectAIOptions",
-            "AIProviderConfig",
-            "StreamAIOptions",
-            "AIChatPluginConfig"
-          ]
-        },
-        {
-          "kind": "added",
-          "summary": "Add one-section text preparation steps for durable workers, with explicit pending/ready outcomes and checkpoint persistence before yielding",
-          "symbols": [
-            "prepareAITextInputStep",
-            "AITextPreparationStep"
-          ]
-        },
-        {
-          "kind": "added",
-          "summary": "Budget final retrieval instructions and tools before compacting text, with matching checkpoint invalidation",
-          "symbols": [
-            "PrepareAITextInputOptions",
-            "prepareAITextInput"
-          ]
-        },
-        {
-          "kind": "fixed",
-          "summary": "Preserve conversation origin across long-text preparation sections, distinguish source facts from embedded instructions, reserve more of the available context for factual notes, and invalidate checkpoints from the previous preparation policy",
-          "symbols": [
-            "prepareAITextInput",
-            "prepareAITextInputStep"
-          ]
-        }
-      ],
-      "date": "2026-09-16",
-      "version": "0.1.0"
-    },
-    {
-      "changes": [
-        {
-          "kind": "added",
-          "summary": "Keep compacted text useful with read-only original-source search, checkpointed ingestion, and capacity-checked terminal tool loops",
-          "symbols": [
-            "createAITextSource",
-            "prepareAITextInput",
-            "PrepareAITextInputOptions",
-            "streamAIWithTools",
-            "StreamAIWithToolsOptions"
-          ]
-        }
-      ],
-      "date": "2026-09-15",
-      "version": "0.0.58"
-    },
-    {
-      "changes": [
-        {
-          "kind": "internal",
-          "summary": "Regenerate the OpenRouter client declaration with equivalent reordered response union types",
-          "symbols": [
-            "createOpenRouterClient"
-          ]
-        },
-        {
-          "kind": "added",
-          "summary": "Add model-specific token capacity, optional provider capacity configuration, and recoverable long-text preparation",
-          "symbols": [
-            "inspectAIInput",
-            "prepareAITextInput",
-            "AIProviderConfig",
-            "OpenAIConfig",
-            "OpenAIResponsesConfig"
-          ]
-        }
-      ],
-      "date": "2026-09-15",
-      "version": "0.0.57"
-    }
-  ]
+	"contract": 1,
+	"name": "@absolutejs/ai",
+	"releases": [
+		{
+			"changes": [
+				{
+					"kind": "fixed",
+					"summary": "Exclude OpenAI cache-write tokens from ordinary input so cache-aware billing never charges the same tokens twice",
+					"symbols": [
+						"openai",
+						"openaiResponses"
+					]
+				}
+			],
+			"date": "2026-09-16",
+			"version": "0.1.3"
+		},
+		{
+			"changes": [
+				{
+					"kind": "fixed",
+					"summary": "Preserve every parallel OpenAI tool result and subsequent conversation turn without overwriting expanded messages",
+					"symbols": [
+						"openai"
+					]
+				}
+			],
+			"date": "2026-09-16",
+			"version": "0.1.2"
+		},
+		{
+			"changes": [
+				{
+					"kind": "added",
+					"summary": "Allow a separate preparation model with its own verified capacity while preserving the final answer model and invalidating notes when the preparation model changes",
+					"symbols": [
+						"PrepareAITextInputOptions",
+						"prepareAITextInput",
+						"prepareAITextInputStep"
+					]
+				},
+				{
+					"kind": "fixed",
+					"summary": "Fill preparation sections with bounded tokenizer-checked refinement, avoid duplicate final counting, and honor cancellation before generation",
+					"symbols": [
+						"prepareAITextInput",
+						"prepareAITextInputStep"
+					]
+				}
+			],
+			"date": "2026-09-16",
+			"version": "0.1.1"
+		},
+		{
+			"changes": [
+				{
+					"kind": "changed",
+					"summary": "Validate model capacity by default across generation, structured output, WebSocket and SSE entry points; reserve token headroom and support one protocol-preserving context recovery attempt. Providers without counting capability now fail explicitly; contextPolicy: false opts into raw behavior.",
+					"symbols": [
+						"generateAI",
+						"generateAIWithTools",
+						"generateObjectAI",
+						"streamAI",
+						"streamAIToSSE",
+						"streamAIWithTools",
+						"inspectAIContext",
+						"createAIStoredToolResultRecovery",
+						"StreamAIWithToolsOptions",
+						"GenerateAIOptions",
+						"GenerateAIResult",
+						"GenerateObjectAIOptions",
+						"AIProviderConfig",
+						"StreamAIOptions",
+						"AIChatPluginConfig"
+					]
+				},
+				{
+					"kind": "added",
+					"summary": "Add one-section text preparation steps for durable workers, with explicit pending/ready outcomes and checkpoint persistence before yielding",
+					"symbols": [
+						"prepareAITextInputStep",
+						"AITextPreparationStep"
+					]
+				},
+				{
+					"kind": "added",
+					"summary": "Budget final retrieval instructions and tools before compacting text, with matching checkpoint invalidation",
+					"symbols": [
+						"PrepareAITextInputOptions",
+						"prepareAITextInput"
+					]
+				},
+				{
+					"kind": "fixed",
+					"summary": "Preserve conversation origin across long-text preparation sections, distinguish source facts from embedded instructions, reserve more of the available context for factual notes, and invalidate checkpoints from the previous preparation policy",
+					"symbols": [
+						"prepareAITextInput",
+						"prepareAITextInputStep"
+					]
+				}
+			],
+			"date": "2026-09-16",
+			"version": "0.1.0"
+		},
+		{
+			"changes": [
+				{
+					"kind": "added",
+					"summary": "Keep compacted text useful with read-only original-source search, checkpointed ingestion, and capacity-checked terminal tool loops",
+					"symbols": [
+						"createAITextSource",
+						"prepareAITextInput",
+						"PrepareAITextInputOptions",
+						"streamAIWithTools",
+						"StreamAIWithToolsOptions"
+					]
+				}
+			],
+			"date": "2026-09-15",
+			"version": "0.0.58"
+		},
+		{
+			"changes": [
+				{
+					"kind": "internal",
+					"summary": "Regenerate the OpenRouter client declaration with equivalent reordered response union types",
+					"symbols": [
+						"createOpenRouterClient"
+					]
+				},
+				{
+					"kind": "added",
+					"summary": "Add model-specific token capacity, optional provider capacity configuration, and recoverable long-text preparation",
+					"symbols": [
+						"inspectAIInput",
+						"prepareAITextInput",
+						"AIProviderConfig",
+						"OpenAIConfig",
+						"OpenAIResponsesConfig"
+					]
+				}
+			],
+			"date": "2026-09-15",
+			"version": "0.0.57"
+		}
+	]
 }

package/dist/ai/index.js

@@ -877,11 +877,12 @@ var extractUsage = (response) => {
   const outputDetails = isRecord(usage.output_tokens_details) ? usage.output_tokens_details : undefined;
   const inputDetails = isRecord(usage.input_tokens_details) ? usage.input_tokens_details : undefined;
   const costDetails = isRecord(usage.cost_details) ? usage.cost_details : undefined;
+  const written = inputDetails && typeof inputDetails.cache_write_tokens === "number" ? inputDetails.cache_write_tokens : 0;
   const normalized = {
     cacheReadInputTokens: cached,
-    cacheWriteInputTokens: inputDetails && typeof inputDetails.cache_write_tokens === "number" ? inputDetails.cache_write_tokens : undefined,
+    cacheWriteInputTokens: written || undefined,
     costCredits: typeof usage.cost === "number" ? usage.cost : undefined,
-    inputTokens: Math.max(0, input - cached),
+    inputTokens: Math.max(0, input - cached - written),
     outputTokens: typeof usage.output_tokens === "number" ? usage.output_tokens : 0,
     reasoningTokens: outputDetails && typeof outputDetails.reasoning_tokens === "number" ? outputDetails.reasoning_tokens : undefined,
     upstreamInferenceCostCredits: costDetails && typeof costDetails.upstream_inference_cost === "number" ? costDetails.upstream_inference_cost : undefined
@@ -1262,30 +1263,6 @@ var buildToolMessages = (blocks) => {
   }
   return messages;
 };
-var processMessageAtIndex = (result, msg, idx) => {
-  if (!hasArrayContent(msg)) {
-    return;
-  }
-  const hasToolBlocks2 = msg.content.some((block) => block.type === "tool_use" || block.type === "tool_result");
-  if (!hasToolBlocks2) {
-    return;
-  }
-  const toolMessages = buildToolMessages(msg.content);
-  result.splice(idx, 1, ...toolMessages);
-};
-var convertSingleMessage = (result, msg, idx) => {
-  if (!msg) {
-    return;
-  }
-  processMessageAtIndex(result, msg, idx);
-};
-var convertToolResultMessages = (messages, params) => {
-  const result = [...messages];
-  for (let idx = 0;idx < params.messages.length; idx++) {
-    convertSingleMessage(result, params.messages[idx], idx);
-  }
-  return result;
-};
 var mapToolDefinitions = (tools) => tools.map((tool) => ({
   function: {
     description: tool.description,
@@ -1342,10 +1319,11 @@ var mapOpenAIContent = (msg) => {
   return msg.content.map(mapContentBlockToOpenAI).filter((mapped) => mapped !== null);
 };
 var buildRequestBody = (params, capabilityModel = params.model) => {
-  const messages = convertToolResultMessages(params.messages.map((msg) => ({
-    content: mapOpenAIContent(msg),
-    role: msg.role
-  })), params);
+  const messages = params.messages.flatMap((msg) => {
+    if (hasArrayContent(msg) && msg.content.some((block) => block.type === "tool_use" || block.type === "tool_result"))
+      return buildToolMessages(msg.content);
+    return [{ content: mapOpenAIContent(msg), role: msg.role }];
+  });
   if (params.systemPrompt) {
     messages.unshift({ content: params.systemPrompt, role: "system" });
   }
@@ -1433,11 +1411,12 @@ var flushPendingToolCalls = function* (pendingToolCalls) {
 var extractUsage2 = (parsedUsage) => {
   const prompt = parsedUsage.prompt_tokens ?? 0;
   const cached = parsedUsage.cached_tokens ?? 0;
+  const written = parsedUsage.cache_write_tokens ?? 0;
   return {
     cacheReadInputTokens: cached,
-    cacheWriteInputTokens: parsedUsage.cache_write_tokens || undefined,
+    cacheWriteInputTokens: written || undefined,
     costCredits: parsedUsage.cost,
-    inputTokens: Math.max(0, prompt - cached),
+    inputTokens: Math.max(0, prompt - cached - written),
     outputTokens: parsedUsage.completion_tokens ?? 0,
     reasoningTokens: parsedUsage.reasoning_tokens || undefined,
     upstreamInferenceCostCredits: parsedUsage.upstream_inference_cost || undefined
@@ -1882,7 +1861,7 @@ var convertMessageContent = (content) => {
   return content.map(mapContentBlock).filter((mapped) => mapped !== null);
 };
 var hasFunctionResponse = (content) => content.some((block) => block.type === "tool_result");
-var convertSingleMessage2 = (msg) => {
+var convertSingleMessage = (msg) => {
   if (typeof msg.content !== "string" && hasFunctionResponse(msg.content)) {
     return { parts: convertMessageContent(msg.content), role: "user" };
   }
@@ -1891,7 +1870,7 @@ var convertSingleMessage2 = (msg) => {
     role: mapRole(msg.role)
   };
 };
-var convertMessages = (messages) => messages.map(convertSingleMessage2);
+var convertMessages = (messages) => messages.map(convertSingleMessage);
 var mapToolDefinitions2 = (tools) => tools.map((tool) => ({
   description: tool.description,
   name: tool.name,
@@ -8727,5 +8706,5 @@ export {
   xai
 };
 
-//# debugId=0E6BB4F28F8F0BB064756E2164756E21
+//# debugId=7535B0C2EE1EFEA864756E2164756E21
 //# sourceMappingURL=index.js.map