ae5b1b5cb5
This adds support for a new flag, `-w,--writable-root`, that can be specified multiple times to _amend_ the list of folders that should be configured as "writable roots" by the sandbox used in `full-auto` mode. Values that are passed as relative paths will be resolved to absolute paths. Incidentally, this required updating a number of the `agent*.test.ts` files: it feels like some of the setup logic across those tests could be consolidated. In my testing, it seems that this might be slightly out of distribution for the model, as I had to explicitly tell it to run `apply_patch` and that it had the permissions to write those files (initially, it just showed me a diff and told me to apply it myself). Nevertheless, I think this is a good starting point.
94 lines
2.5 KiB
TypeScript
94 lines
2.5 KiB
TypeScript
import { describe, it, expect, vi } from "vitest";
|
||
|
||
// ---------------------------------------------------------------------------
|
||
// Mock helpers
|
||
// ---------------------------------------------------------------------------
|
||
|
||
const openAiState: { createSpy?: ReturnType<typeof vi.fn> } = {};
|
||
|
||
vi.mock("openai", () => {
|
||
class FakeOpenAI {
|
||
public responses = {
|
||
create: (...args: Array<any>) => openAiState.createSpy!(...args),
|
||
};
|
||
}
|
||
|
||
class APIConnectionTimeoutError extends Error {}
|
||
|
||
return {
|
||
__esModule: true,
|
||
default: FakeOpenAI,
|
||
APIConnectionTimeoutError,
|
||
};
|
||
});
|
||
|
||
vi.mock("../src/approvals.js", () => ({
|
||
__esModule: true,
|
||
alwaysApprovedCommands: new Set<string>(),
|
||
canAutoApprove: () => ({ type: "auto-approve", runInSandbox: false } as any),
|
||
isSafeCommand: () => null,
|
||
}));
|
||
|
||
vi.mock("../src/format-command.js", () => ({
|
||
__esModule: true,
|
||
formatCommandForDisplay: (c: Array<string>) => c.join(" "),
|
||
}));
|
||
|
||
vi.mock("../src/utils/agent/log.js", () => ({
|
||
__esModule: true,
|
||
log: () => {},
|
||
isLoggingEnabled: () => false,
|
||
}));
|
||
|
||
import { AgentLoop } from "../src/utils/agent/agent-loop.js";
|
||
|
||
describe("AgentLoop – max_tokens too large error", () => {
|
||
it("shows context‑length system message and resolves", async () => {
|
||
const err: any = new Error(
|
||
"max_tokens is too large: 167888. This model supports at most 100000 completion tokens, whereas you provided 167888.",
|
||
);
|
||
err.type = "invalid_request_error";
|
||
err.param = "max_tokens";
|
||
err.status = 400;
|
||
|
||
openAiState.createSpy = vi.fn(async () => {
|
||
throw err;
|
||
});
|
||
|
||
const received: Array<any> = [];
|
||
|
||
const agent = new AgentLoop({
|
||
additionalWritableRoots: [],
|
||
model: "any",
|
||
instructions: "",
|
||
approvalPolicy: { mode: "auto" } as any,
|
||
onItem: (i) => received.push(i),
|
||
onLoading: () => {},
|
||
getCommandConfirmation: async () => ({ review: "yes" } as any),
|
||
onLastResponseId: () => {},
|
||
});
|
||
|
||
const userMsg = [
|
||
{
|
||
type: "message",
|
||
role: "user",
|
||
content: [{ type: "input_text", text: "hello" }],
|
||
},
|
||
];
|
||
|
||
await expect(agent.run(userMsg as any)).resolves.not.toThrow();
|
||
|
||
// allow asynchronous onItem calls to flush
|
||
await new Promise((r) => setTimeout(r, 20));
|
||
|
||
const sysMsg = received.find(
|
||
(i) =>
|
||
i.role === "system" &&
|
||
typeof i.content?.[0]?.text === "string" &&
|
||
i.content[0].text.includes("exceeds the maximum context length"),
|
||
);
|
||
|
||
expect(sysMsg).toBeTruthy();
|
||
});
|
||
});
|