@inference-gateway/sdk 0.7.2 → 0.8.2

package/CHANGELOG.md

@@ -2,6 +2,69 @@
 
 All notable changes to this project will be documented in this file.
 
+## [0.8.2](https://github.com/inference-gateway/typescript-sdk/compare/v0.8.1...v0.8.2) (2026-05-06)
+
+### 👷 CI
+
+* Grant id-token write permission for npm Trusted Publishing ([36b2e88](https://github.com/inference-gateway/typescript-sdk/commit/36b2e881313b4b5d73ba6479eb5c7e664d4dd291))
+
+## [0.8.1](https://github.com/inference-gateway/typescript-sdk/compare/v0.8.0...v0.8.1) (2026-05-06)
+
+### 🐛 Bug Fixes
+
+* Exclude test files from published npm tarball ([56a3f5e](https://github.com/inference-gateway/typescript-sdk/commit/56a3f5e2caeb15baa034e1b35d04f92a56ca96f2)), closes [package.json#files](https://github.com/inference-gateway/package.json/issues/files)
+
+## [0.8.0](https://github.com/inference-gateway/typescript-sdk/compare/v0.7.3...v0.8.0) (2026-05-06)
+
+### ✨ Features
+
+* Adopt latest OpenAPI spec and preserve tool-call extra_content ([193184c](https://github.com/inference-gateway/typescript-sdk/commit/193184c5d9513fc8681606da860c409655f9453f))
+* **providers:** Add Google provider support ([#21](https://github.com/inference-gateway/typescript-sdk/issues/21)) ([a6381d7](https://github.com/inference-gateway/typescript-sdk/commit/a6381d7357d22f32434599e204955fff4db2e3e0)), closes [#19](https://github.com/inference-gateway/typescript-sdk/issues/19)
+
+### ♻️ Improvements
+
+* Rename all instances of deepseek-chat to deepseek-v4-flash ([465851e](https://github.com/inference-gateway/typescript-sdk/commit/465851e8f49fba153d1ded6add5d3511507ee50e))
+
+### 👷 CI
+
+* Add Claude Code GitHub Workflow ([#20](https://github.com/inference-gateway/typescript-sdk/issues/20)) ([83e61a0](https://github.com/inference-gateway/typescript-sdk/commit/83e61a01fb2fcfc63f06ea2fc1cebb6076285b17))
+* Bump all actions to latest ([d23eef1](https://github.com/inference-gateway/typescript-sdk/commit/d23eef13fa6f4d015059b98bd90a75c4b8f3ddf7))
+
+### 📚 Documentation
+
+*  Add more examples how to use this sdk ([#16](https://github.com/inference-gateway/typescript-sdk/issues/16)) ([5bddd0b](https://github.com/inference-gateway/typescript-sdk/commit/5bddd0beb693e1ed3341f8c48511dd5e9045729d))
+* Add CLAUDE.md for project guidance and development instructions ([47645bf](https://github.com/inference-gateway/typescript-sdk/commit/47645bfd0f9e0f5c0c83051acbdd52253318522d))
+
+### 🔧 Miscellaneous
+
+* Add .vscode to gitignore ([6e3117e](https://github.com/inference-gateway/typescript-sdk/commit/6e3117e0b6344747acc36ba21f747d10118560a2))
+* Add issue templates ([769f017](https://github.com/inference-gateway/typescript-sdk/commit/769f017bf6810c687bc11795fbcb545ffaaaa446))
+* Add project configuration and documentation ([f4bde02](https://github.com/inference-gateway/typescript-sdk/commit/f4bde02757e2d32512ec535e254217d3d81d76a0))
+* Delete CLAUDE.md ([56bc06c](https://github.com/inference-gateway/typescript-sdk/commit/56bc06c85b061d6c7674173b60ab3dde6e0f9b69))
+* **deps:** Bump all version to latest ([0d02bc5](https://github.com/inference-gateway/typescript-sdk/commit/0d02bc5045696a3046cb5460eeb1601b039accdb))
+* **deps:** Install task runner for local env ([be7a77f](https://github.com/inference-gateway/typescript-sdk/commit/be7a77ffe5bcb05980a73604190661b2a079cb2e))
+* **deps:** Update to their latest ([741971b](https://github.com/inference-gateway/typescript-sdk/commit/741971bd3d9f8a4ab80590b5c3b6a374dc292940))
+* Download the latest oas ([9fcf5da](https://github.com/inference-gateway/typescript-sdk/commit/9fcf5da1b7cc12661e3222155c885fae77cb1931))
+* Lock the versions of npm and node in package.json ([9690d75](https://github.com/inference-gateway/typescript-sdk/commit/9690d7501b8edd49bb30c9d0274247802c9bfde4))
+* Remove deprecated lines from husky ([177d464](https://github.com/inference-gateway/typescript-sdk/commit/177d46459c9ffb190dbad41808e7f9ed207ed1e7))
+* Replace devcontainer with Flox environment and streamline CI ([08c1eaf](https://github.com/inference-gateway/typescript-sdk/commit/08c1eaf3ac525153445ab2e1db5fc7b0c79c9457))
+* Run task generate-types ([e01feff](https://github.com/inference-gateway/typescript-sdk/commit/e01feffba970085449bae28fcc49132f84e27ae8))
+* Update GitHub Actions dependencies ([31006d8](https://github.com/inference-gateway/typescript-sdk/commit/31006d898f64d996d225818f3b93f38df644ae3f))
+
+### 🎨 Miscellaneous
+
+* Fix markdown lint errors across all documentation ([419062d](https://github.com/inference-gateway/typescript-sdk/commit/419062d726106aa84b88f805a75f3acb48582ca7))
+
+## [0.7.3](https://github.com/inference-gateway/typescript-sdk/compare/v0.7.2...v0.7.3) (2025-06-01)
+
+### ♻️ Improvements
+
+* Enhance stream processing with abort signal support and increase default timeout ([#18](https://github.com/inference-gateway/typescript-sdk/issues/18)) ([3778138](https://github.com/inference-gateway/typescript-sdk/commit/377813851b6635ca7aafe2a5c9888b720736c9f5))
+
+### 🔧 Miscellaneous
+
+* Update MCP example README and remove unused example file ([99b34e7](https://github.com/inference-gateway/typescript-sdk/commit/99b34e70edf0c8aada1d0e0d0874481ea8381a79))
+
 ## [0.7.2](https://github.com/inference-gateway/typescript-sdk/compare/v0.7.1...v0.7.2) (2025-05-30)
 
 ### 📚 Documentation

package/README.md

@@ -62,7 +62,8 @@ try {
 
 ### Listing MCP Tools
 
-To list available Model Context Protocol (MCP) tools (only available when EXPOSE_MCP is enabled):
+To list available Model Context Protocol (MCP) tools (only available when
+EXPOSE_MCP is enabled):
 
 ```typescript
 import { InferenceGatewayClient } from '@inference-gateway/sdk';
@@ -116,7 +117,7 @@ try {
         },
       ],
     },
-    Provider.OpenAI
+    Provider.openai
   ); // Provider is optional
 
   console.log('Response:', response.choices[0].message.content);
@@ -159,7 +160,7 @@ try {
       onFinish: () => console.log('\nStream completed'),
       onError: (error) => console.error('Stream error:', error),
     },
-    Provider.Groq // Provider is optional
+    Provider.groq // Provider is optional
   );
 } catch (error) {
   console.error('Error:', error);
@@ -241,7 +242,7 @@ const client = new InferenceGatewayClient({
 });
 
 try {
-  const response = await client.proxy(Provider.OpenAI, 'embeddings', {
+  const response = await client.proxy(Provider.openai, 'embeddings', {
     method: 'POST',
     body: JSON.stringify({
       model: 'text-embedding-ada-002',
@@ -300,7 +301,8 @@ For more examples, check the [examples directory](./examples).
 
 ## Contributing
 
-Please refer to the [CONTRIBUTING.md](CONTRIBUTING.md) file for information about how to get involved. We welcome issues, questions, and pull requests.
+Please refer to the [CONTRIBUTING.md](CONTRIBUTING.md) file for information
+about how to get involved. We welcome issues, questions, and pull requests.
 
 ## License
 

package/dist/src/client.d.ts

@@ -54,8 +54,9 @@ export declare class InferenceGatewayClient {
      * @param request - Chat completion request (must include at least model and messages)
      * @param callbacks - Callbacks for handling streaming events
      * @param provider - Optional provider to use for this request
+     * @param abortSignal - Optional AbortSignal to cancel the request
      */
-    streamChatCompletion(request: Omit<SchemaCreateChatCompletionRequest, 'stream' | 'stream_options'>, callbacks: ChatCompletionStreamCallbacks, provider?: Provider): Promise<void>;
+    streamChatCompletion(request: Omit<SchemaCreateChatCompletionRequest, 'stream' | 'stream_options'>, callbacks: ChatCompletionStreamCallbacks, provider?: Provider, abortSignal?: AbortSignal): Promise<void>;
     /**
      * Initiates a streaming request to the chat completions endpoint
      */

package/dist/src/client.js

@@ -13,12 +13,15 @@ class StreamProcessor {
         this.callbacks = callbacks;
         this.clientProvidedTools = clientProvidedTools;
     }
-    async processStream(body) {
+    async processStream(body, abortSignal) {
         const reader = body.getReader();
         const decoder = new TextDecoder();
         let buffer = '';
         try {
             while (true) {
+                if (abortSignal?.aborted) {
+                    throw new Error('Stream processing was aborted');
+                }
                 const { done, value } = await reader.read();
                 if (done)
                     break;
@@ -34,6 +37,10 @@ class StreamProcessor {
             }
         }
         catch (error) {
+            if (abortSignal?.aborted || error.name === 'AbortError') {
+                console.log('Stream processing was cancelled');
+                return;
+            }
             const apiError = {
                 error: error.message || 'Unknown error',
             };
@@ -126,6 +133,7 @@ class StreamProcessor {
                         name: toolCallChunk.function?.name || '',
                         arguments: toolCallChunk.function?.arguments || '',
                     },
+                    extra_content: toolCallChunk.extra_content,
                 });
             }
             else {
@@ -140,6 +148,9 @@ class StreamProcessor {
                     existingToolCall.function.arguments +=
                         toolCallChunk.function.arguments;
                 }
+                if (toolCallChunk.extra_content) {
+                    existingToolCall.extra_content = toolCallChunk.extra_content;
+                }
             }
         }
     }
@@ -150,10 +161,10 @@ class StreamProcessor {
         }
     }
     finalizeIncompleteToolCalls() {
-        for (const [, toolCall] of this.incompleteToolCalls.entries()) {
+        this.incompleteToolCalls.forEach((toolCall) => {
             if (!toolCall.id || !toolCall.function.name) {
                 globalThis.console.warn('Incomplete tool call detected:', toolCall);
-                continue;
+                return;
             }
             const completedToolCall = {
                 id: toolCall.id,
@@ -162,6 +173,9 @@ class StreamProcessor {
                     name: toolCall.function.name,
                     arguments: toolCall.function.arguments,
                 },
+                ...(toolCall.extra_content && {
+                    extra_content: toolCall.extra_content,
+                }),
             };
             if (this.isMCPTool(toolCall.function.name)) {
                 try {
@@ -171,13 +185,20 @@ class StreamProcessor {
                     this.callbacks.onMCPTool?.(completedToolCall);
                 }
                 catch (argError) {
-                    globalThis.console.warn(`Invalid MCP tool arguments for ${toolCall.function.name}:`, argError);
+                    const isIncompleteJSON = toolCall.function.arguments &&
+                        !toolCall.function.arguments.trim().endsWith('}');
+                    if (isIncompleteJSON) {
+                        globalThis.console.warn(`Incomplete MCP tool arguments for ${toolCall.function.name} (stream was likely interrupted):`, toolCall.function.arguments);
+                    }
+                    else {
+                        globalThis.console.warn(`Invalid MCP tool arguments for ${toolCall.function.name}:`, argError);
+                    }
                 }
             }
             else {
                 this.callbacks.onTool?.(completedToolCall);
             }
-        }
+        });
         this.incompleteToolCalls.clear();
     }
     isMCPTool(toolName) {
@@ -199,7 +220,7 @@ class InferenceGatewayClient {
         this.apiKey = options.apiKey;
         this.defaultHeaders = options.defaultHeaders || {};
         this.defaultQuery = options.defaultQuery || {};
-        this.timeout = options.timeout || 30000;
+        this.timeout = options.timeout || 60000; // Increased default timeout to 60 seconds
         this.fetchFn = options.fetch || globalThis.fetch;
     }
     /**
@@ -291,10 +312,11 @@ class InferenceGatewayClient {
      * @param request - Chat completion request (must include at least model and messages)
      * @param callbacks - Callbacks for handling streaming events
      * @param provider - Optional provider to use for this request
+     * @param abortSignal - Optional AbortSignal to cancel the request
      */
-    async streamChatCompletion(request, callbacks, provider) {
+    async streamChatCompletion(request, callbacks, provider, abortSignal) {
         try {
-            const response = await this.initiateStreamingRequest(request, provider);
+            const response = await this.initiateStreamingRequest(request, provider, abortSignal);
             if (!response.body) {
                 const error = {
                     error: 'Response body is not readable',
@@ -313,7 +335,7 @@ class InferenceGatewayClient {
                 }
             }
             const streamProcessor = new StreamProcessor(callbacks, clientProvidedTools);
-            await streamProcessor.processStream(response.body);
+            await streamProcessor.processStream(response.body, abortSignal);
         }
         catch (error) {
             const apiError = {
@@ -326,7 +348,7 @@ class InferenceGatewayClient {
     /**
      * Initiates a streaming request to the chat completions endpoint
      */
-    async initiateStreamingRequest(request, provider) {
+    async initiateStreamingRequest(request, provider, abortSignal) {
         const query = {};
         if (provider) {
             query.provider = provider;
@@ -345,6 +367,9 @@ class InferenceGatewayClient {
             headers.set('Authorization', `Bearer ${this.apiKey}`);
         }
         const controller = new AbortController();
+        const combinedSignal = abortSignal
+            ? AbortSignal.any([abortSignal, controller.signal])
+            : controller.signal;
         const timeoutId = globalThis.setTimeout(() => controller.abort(), this.timeout);
         try {
             const response = await this.fetchFn(url, {
@@ -357,7 +382,7 @@ class InferenceGatewayClient {
                         include_usage: true,
                     },
                 }),
-                signal: controller.signal,
+                signal: combinedSignal,
             });
             if (!response.ok) {
                 let errorMessage = `HTTP error! status: ${response.status}`;