Never stop coding. Free MIT AI gateway: one endpoint, 375 providers (150+ free), 1,400+ models Kimi, Claude, GPT, Gemini, GLM, DeepSeek, MiniMax. Works with Claude Code, Codex, Cursor, OpenCode, Cline & Copilot. Quota-aware auto-fallback, RTK+Caveman compression saves 15-95% tokens, MCP/A2A, Desktop/PWA. Built by 900+ contributors
TypeScript a2aai-agentsai-gatewayanthropicclaudeclaude-codeclinecodexcopilotcursordeepseekfree-aigeminikimillm-gatewaymcpopenaiopenai-proxyqwentoken-saver
/**
* Connect-RPC ends a server stream with an end-of-stream frame (flag 0x02)
* whose payload is JSON — `{}` on success, `{"error":{...}}` on failure — not
* protobuf. Treating it as a malformed protobuf frame failed every such turn
* with "cursor-agent frame decode failed (flag=2, size=2)".
* Run: node --import tsx/esm --test tests/unit/cursor-stream-driver-end-stream.test.ts
*/
import assert from "node:assert/strict";
import { EventEmitter } from "node:events";
import test from "node:test";
import { driveCursorH2 } from "../../open-sse/executors/cursor/streamDriver.ts";
function frame(flag: number, payload: Buffer): Buffer {
const header = Buffer.alloc(5);
header[0] = flag;
header.writeUInt32BE(payload.length, 1);
return Buffer.concat([header, payload]);
}
function fakeH2() {
const req = Object.assign(new EventEmitter(), {
close: () => {},
resume: () => {},
pause: () => {},
}) as unknown as import("node:http2").ClientHttp2Stream;
const client = { close: () => {} } as unknown as import("node:http2").ClientHttp2Session;
return { req, client };
}
type Ctx = {
model: string;
emittedRoleChunk: boolean;
endReason: "turn_ended" | "kv_after_text" | "tool_calls" | "server_end" | null;
leftoverBytes: Buffer;
unknownExecField: number | null;
};
function drive(initialBytes: Buffer, onFrame: (payload: Buffer) => void = () => {}) {
const { req, client } = fakeH2();
const ctx: Ctx = {
model: "grok-4.6",
emittedRoleChunk: false,
endReason: null,
leftoverBytes: Buffer.alloc(0),
unknownExecField: null,
};
const task = driveCursorH2({ req, client, initialBytes }, ctx as never, {
debugEnabled: false,
debugLog: () => {},
onFrame: (payload) => onFrame(payload),
});
// The server closes the stream right after its end-of-stream frame.
setImmediate(() => req.emit("end"));
return { task, ctx };
}
test("a successful end-of-stream frame ends the turn instead of failing it", async () => {
const seen: Buffer[] = [];
const { task, ctx } = drive(frame(0x02, Buffer.from("{}")), (p) => seen.push(p));
await task;
assert.equal(ctx.endReason, "server_end");
assert.equal(seen.length, 0, "the JSON trailer must not reach the protobuf frame handler");
});
test("frames before the end-of-stream frame are still handled", async () => {
const seen: Buffer[] = [];
const data = Buffer.from([0x0a, 0x00]);
const { task } = drive(Buffer.concat([frame(0x00, data), frame(0x02, Buffer.from("{}"))]), (p) =>
seen.push(p)
);
await task;
assert.deepEqual(seen, [data]);
});
test("an end-of-stream error is reported with its Connect code", async () => {
const trailer = JSON.stringify({ error: { code: "resource_exhausted", message: "slow down" } });
const { task } = drive(frame(0x02, Buffer.from(trailer)));
await assert.rejects(task, /resource_exhausted/);
});