Merge branch 'v1' into cb/context-editing-middleware

Author: christian-bromann

examples/src/createAgent/middleware/toolCallLimit.ts

@@ -0,0 +1,99 @@
+/**
+ * Basic example demonstrating tool call limit middleware.
+ *
+ * This middleware helps prevent infinite loops or excessive tool usage
+ * by limiting the number of tool calls an agent can make.
+ */
+
+import { z } from "zod";
+import { createAgent, tool, toolCallLimitMiddleware } from "langchain";
+import { ChatOpenAI } from "@langchain/openai";
+import { HumanMessage } from "@langchain/core/messages";
+import { MemorySaver } from "@langchain/langgraph";
+const config = { configurable: { thread_id: "demo-thread" } };
+
+/**
+ * Define a simple search tool
+ */
+const searchTool = tool(
+  async ({ query }) => {
+    console.log(`Searching for: ${query}`);
+    return `Results for: ${query}`;
+  },
+  {
+    name: "search",
+    description: "Search for information",
+    schema: z.object({
+      query: z.string(),
+    }),
+  }
+);
+
+/**
+ * Create an agent with a tool call limit
+ */
+const agent = createAgent({
+  model: new ChatOpenAI({ model: "gpt-4o-mini" }),
+  tools: [searchTool],
+  checkpointer: new MemorySaver(),
+  middleware: [
+    /**
+     * Limit to 3 tool calls per conversation
+     */
+    toolCallLimitMiddleware({
+      threadLimit: 3,
+      /**
+       * Gracefully end when limit is reached
+       */
+      exitBehavior: "end",
+    }),
+  ],
+});
+
+/**
+ * Example conversation that would exceed the limit
+ */
+const result = await agent.invoke(
+  {
+    messages: [
+      new HumanMessage(
+        "Search for 'AI', 'ML', 'Deep Learning', 'Neural Networks', and 'LLMs'"
+      ),
+    ],
+  },
+  { configurable: { thread_id: "demo-thread" } }
+);
+
+console.log("\nAgent response:");
+console.log(result.messages[result.messages.length - 1].content);
+
+/**
+ * Create an agent with a tool call limit
+ */
+const agent2 = createAgent({
+  model: new ChatOpenAI({ model: "gpt-4o-mini" }),
+  tools: [searchTool],
+  checkpointer: new MemorySaver(),
+  middleware: [
+    /**
+     * Limit to 3 tool calls per conversation
+     */
+    toolCallLimitMiddleware({
+      threadLimit: 3,
+      /**
+       * Gracefully end when limit is reached
+       */
+      exitBehavior: "end",
+    }),
+  ],
+});
+
+const result2 = await agent.invoke(
+  {
+    messages: [new HumanMessage("Search for 'AI' and 'ML'")],
+  },
+  { configurable: { thread_id: "demo-thread" } }
+);
+
+console.log("\nAgent response:");
+console.log(result2.messages[result2.messages.length - 1].content);

libs/langchain/src/agents/middlewareAgent/middleware/index.ts

@@ -33,6 +33,11 @@ export {
   type ClearToolUsesEditConfig,
   type TokenCounter,
 } from "./contextEditing.js";
+export {
+  toolCallLimitMiddleware,
+  ToolCallLimitExceededError,
+  type ToolCallLimitConfig,
+} from "./toolCallLimit.js";
 export {
   modelCallLimitMiddleware,
   type ModelCallLimitMiddlewareConfig,