fix(api): send Content-Location header for stateless runs (#1512)

.changeset/brave-rings-mate.md

@@ -0,0 +1,5 @@
+---
+"@langchain/langgraph-api": patch
+---
+
+fix(api): send Content-Location header for stateless runs

libs/langgraph-api/src/api/runs.mts

@@ -205,6 +205,7 @@ api.post("/runs/stream", zValidator("json", schemas.RunCreate), async (c) => {
     headers: c.req.raw.headers,
   });
 
+  c.header("Content-Location", `/runs/${run.run_id}`);
   return streamSSE(c, async (stream) => {
     const cancelOnDisconnect =
       payload.on_disconnect === "cancel"
@@ -243,7 +244,7 @@ api.get(
     const query = c.req.valid("query");
 
     const lastEventId = c.req.header("Last-Event-ID") || undefined;
-
+    c.header("Content-Location", `/runs/${run_id}`);
     return streamSSE(c, async (stream) => {
       const cancelOnDisconnect = query.cancel_on_disconnect
         ? getDisconnectAbortSignal(c, stream)
@@ -272,6 +273,7 @@ api.post("/runs/wait", zValidator("json", schemas.RunCreate), async (c) => {
     auth: c.var.auth,
     headers: c.req.raw.headers,
   });
+  c.header("Content-Location", `/runs/${run.run_id}`);
   return waitKeepAlive(c, Runs.wait(run.run_id, undefined, c.var.auth));
 });
 
@@ -282,6 +284,7 @@ api.post("/runs", zValidator("json", schemas.RunCreate), async (c) => {
     auth: c.var.auth,
     headers: c.req.raw.headers,
   });
+  c.header("Content-Location", `/runs/${run.run_id}`);
   return jsonExtra(c, run);
 });
 

libs/langgraph-api/tests/api.test.mts

@@ -9,7 +9,15 @@ import {
 } from "@langchain/langgraph-sdk";
 import { RemoteGraph } from "@langchain/langgraph/remote";
 import postgres from "postgres";
-import { afterAll, beforeAll, beforeEach, describe, expect, it } from "vitest";
+import {
+  afterAll,
+  beforeAll,
+  beforeEach,
+  describe,
+  expect,
+  it,
+  vi,
+} from "vitest";
 import { findLast, gatherIterator, truncate } from "./utils.mjs";
 import { type ChildProcess, spawn } from "node:child_process";
 import { randomUUID } from "node:crypto";
@@ -2812,12 +2820,17 @@ describe("runtime API", () => {
     });
     const thread = await client.threads.create();
 
+    const onRunCreated = vi.fn();
+
     expect(
       await client.runs.wait(thread.thread_id, assistant.assistant_id, {
         input: { messages: [{ role: "human", content: "input" }] },
         context: { model: "openai" },
+        onRunCreated,
       })
     ).toEqual({ model: "openai" });
+
+    expect(onRunCreated).toHaveBeenCalledTimes(1);
   });
 
   it("schema", async () => {
@@ -2851,11 +2864,14 @@ describe("runtime API", () => {
       context: { model: "anthropic" },
     });
 
+    const onRunCreated = vi.fn();
     expect(
       await client.runs.wait(null, assistant.assistant_id, {
         input: { messages: [{ role: "human", content: "input" }] },
+        onRunCreated,
       })
     ).toEqual({ model: "anthropic" });
+    expect(onRunCreated).toHaveBeenCalledTimes(1);
 
     // patch to say openai
     await client.assistants.update(assistant.assistant_id, {
@@ -2865,15 +2881,20 @@ describe("runtime API", () => {
     expect(
       await client.runs.wait(null, assistant.assistant_id, {
         input: { messages: [{ role: "human", content: "input" }] },
+        onRunCreated,
       })
     ).toEqual({ model: "openai" });
+    expect(onRunCreated).toHaveBeenCalledTimes(2);
   });
 
   it("default from env", async () => {
+    const onRunCreated = vi.fn();
     expect(
       await client.runs.wait(null, "simple_runtime", {
         input: { messages: [{ role: "human", content: "input" }] },
+        onRunCreated,
       })
     ).toEqual({ model: "unknown" });
+    expect(onRunCreated).toHaveBeenCalledTimes(1);
   });
 });