The GPT-account (Codex) path supports tool calls and worked before; my forced tool_choice broke it and was the wrong fix. The real regression was strict:null in the rebuild's tool definitions, which made the endpoint drop the tool list so the model had nothing to call — a "Done" run that hired no one. Forcing then turned that silent failure into a red one. Restore the original "auto" loop (no forced tool_choice, no finish tool) and keep only the one-line strict:null omission. Codex now receives valid tools on the working "auto" choice, as in the version that worked. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
176 lines
9.3 KiB
TypeScript
176 lines
9.3 KiB
TypeScript
import { existsSync, mkdirSync, readFileSync } from "node:fs";
|
|
import { randomUUID } from "node:crypto";
|
|
import { tmpdir } from "node:os";
|
|
import { join } from "node:path";
|
|
import { describe, expect, test } from "vitest";
|
|
import { DatabaseConnection } from "../../src/main/database/connection";
|
|
import { OrganizationService } from "../../src/main/services/organization-service";
|
|
import { ChatService } from "../../src/main/services/chat-service";
|
|
import { runAgentLoop } from "../../src/main/services/agent-run-loop";
|
|
import type { AgentToolContext } from "../../src/main/services/agent-tools";
|
|
import type { AgentTurnResult, ToolCall } from "../../src/main/services/provider-tool-turn";
|
|
|
|
interface ScriptedTurn {
|
|
text: string;
|
|
toolCalls: ToolCall[];
|
|
}
|
|
|
|
function fakeProvider(script: ScriptedTurn[]) {
|
|
let index = 0;
|
|
return {
|
|
runToolTurn: async (): Promise<AgentTurnResult & { requestId: string }> => {
|
|
const turn = script[index] ?? { text: "done", toolCalls: [] };
|
|
index += 1;
|
|
return { text: turn.text, toolCalls: turn.toolCalls, rawResponseJson: "{}", requestId: `req_${index}` };
|
|
},
|
|
};
|
|
}
|
|
|
|
function call(name: string, args: Record<string, unknown>): ToolCall {
|
|
return { id: `call_${randomUUID()}`, name, arguments: args };
|
|
}
|
|
|
|
describe("agent run loop (real tool dispatch)", () => {
|
|
test("the CEO hires, creates a task, writes a deliverable, then converges", async () => {
|
|
const connection = new DatabaseConnection(join(tmpdir(), `ac-loop-${randomUUID()}.db`));
|
|
const organization = new OrganizationService(connection.db);
|
|
const workspaceRoot = join(tmpdir(), `ac-ws-${randomUUID()}`);
|
|
mkdirSync(workspaceRoot, { recursive: true });
|
|
|
|
let orgId = "";
|
|
let ceoId = "";
|
|
let providerId = "";
|
|
const workspaceId = "wsp_test";
|
|
connection.transaction(() => {
|
|
orgId = organization.createOrganization({
|
|
organizationName: "OrbitOps", employeeName: "Yang", mission: "Run an AI company.", goal: "Win customers.",
|
|
});
|
|
providerId = (connection.db.prepare("select id from providers where org_id = ? and provider_key = 'openai'").get(orgId) as { id: string }).id;
|
|
connection.db.prepare(`
|
|
insert into workspaces(id, org_id, name, local_path, repo_url, repo_ref, project_id, is_primary, created_at, updated_at)
|
|
values (?, ?, 'Default Workspace', ?, '', '', null, 1, datetime('now'), datetime('now'))
|
|
`).run(workspaceId, orgId, workspaceRoot);
|
|
ceoId = organization.createAgent({
|
|
orgId, name: "CEO", role: "CEO", title: "Chief Executive Officer", department: "Executive",
|
|
modelProviderId: providerId, modelId: "gpt-5.4", budgetCents: 25000, managerAgentId: null,
|
|
roleLevel: "c_suite", workspaceId,
|
|
});
|
|
});
|
|
|
|
const toolContext: AgentToolContext = {
|
|
orgId,
|
|
agent: { id: ceoId, modelProviderId: providerId, modelId: "gpt-5.4" },
|
|
taskProjectId: null,
|
|
taskGoalId: null,
|
|
workspace: { id: workspaceId, localPath: workspaceRoot },
|
|
};
|
|
|
|
const script: ScriptedTurn[] = [
|
|
{ text: "Staffing the team.", toolCalls: [
|
|
call("hire_agent", { name: "Alex Rivera", role: "Growth Marketer", department: "Marketing" }),
|
|
call("hire_agent", { name: "Sam Lee", role: "Operations Lead", department: "Operations" }),
|
|
] },
|
|
{ text: "Delegating.", toolCalls: [
|
|
call("create_task", { title: "Build the booking site", description: "Ship a landing + booking flow.", assignee_agent_name: "Alex Rivera" }),
|
|
] },
|
|
{ text: "Writing the plan.", toolCalls: [
|
|
call("write_deliverable", { path: "plan/operating-plan.md", content: "# Operating Plan\n\n1. Validate demand.\n2. Launch booking site.\n" }),
|
|
call("send_message", { channel: "general", body: "Team staffed and the operating plan is live. Booking site is next." }),
|
|
] },
|
|
{ text: "Team staffed, work delegated, plan written.", toolCalls: [] },
|
|
];
|
|
|
|
let toolResultNotifications = 0;
|
|
const result = await runAgentLoop(
|
|
{
|
|
providers: fakeProvider(script),
|
|
tools: { db: connection.db, organization, chat: new ChatService(connection.db) },
|
|
onToolResults: () => { toolResultNotifications += 1; },
|
|
},
|
|
{ orgId, providerId, workspaceId, instructions: "You are the CEO.", initialUserTurn: "Start the company.", toolContext },
|
|
);
|
|
|
|
expect(result.iterations).toBe(4);
|
|
expect(result.toolCallCount).toBe(5);
|
|
expect(result.finalText).toContain("Team staffed");
|
|
// The UI is told about new data after every tool-bearing iteration, not only at the end.
|
|
expect(toolResultNotifications).toBe(3);
|
|
|
|
const agentNames = (connection.db.prepare("select name from agents where org_id = ? order by name").all(orgId) as Array<{ name: string }>).map((r) => r.name);
|
|
expect(agentNames).toEqual(["Alex Rivera", "CEO", "Sam Lee"]);
|
|
|
|
const task = connection.db.prepare("select title, assignee_agent_id from tasks where org_id = ?").get(orgId) as { title: string; assignee_agent_id: string } | undefined;
|
|
expect(task?.title).toBe("Build the booking site");
|
|
const alexId = (connection.db.prepare("select id from agents where org_id = ? and name = 'Alex Rivera'").get(orgId) as { id: string }).id;
|
|
expect(task?.assignee_agent_id).toBe(alexId);
|
|
|
|
const deliverablePath = join(workspaceRoot, "plan", "operating-plan.md");
|
|
expect(existsSync(deliverablePath)).toBe(true);
|
|
expect(readFileSync(deliverablePath, "utf8")).toContain("Operating Plan");
|
|
|
|
// The CEO's coordination message lands in #general so the channel is no longer blank.
|
|
const generalSpaceId = (connection.db.prepare("select id from chat_spaces where org_id = ? and name = 'general'").get(orgId) as { id: string }).id;
|
|
const messages = connection.db.prepare(
|
|
"select sender_type, sender_id, body_md, space_id from chat_messages where org_id = ?",
|
|
).all(orgId) as Array<{ sender_type: string; sender_id: string; body_md: string; space_id: string }>;
|
|
expect(messages).toHaveLength(1);
|
|
expect(messages[0]).toMatchObject({ sender_type: "agent", sender_id: ceoId, space_id: generalSpaceId });
|
|
expect(messages[0]?.body_md).toContain("operating plan is live");
|
|
|
|
connection.close();
|
|
});
|
|
|
|
test("rejects deliverable paths that escape the workspace", async () => {
|
|
const connection = new DatabaseConnection(join(tmpdir(), `ac-loop2-${randomUUID()}.db`));
|
|
const organization = new OrganizationService(connection.db);
|
|
const workspaceRoot = join(tmpdir(), `ac-ws2-${randomUUID()}`);
|
|
mkdirSync(workspaceRoot, { recursive: true });
|
|
let orgId = "";
|
|
let ceoId = "";
|
|
let providerId = "";
|
|
connection.transaction(() => {
|
|
orgId = organization.createOrganization({ organizationName: "OrbitOps", employeeName: "Yang", mission: "m", goal: "g" });
|
|
providerId = (connection.db.prepare("select id from providers where org_id = ? and provider_key = 'openai'").get(orgId) as { id: string }).id;
|
|
ceoId = organization.createAgent({ orgId, name: "CEO", role: "CEO", department: "Executive", modelProviderId: providerId, modelId: "gpt-5.4", budgetCents: 25000, managerAgentId: null });
|
|
});
|
|
const ctx: AgentToolContext = {
|
|
orgId, agent: { id: ceoId, modelProviderId: providerId, modelId: "gpt-5.4" },
|
|
taskProjectId: null, taskGoalId: null, workspace: { id: "wsp", localPath: workspaceRoot },
|
|
};
|
|
const script: ScriptedTurn[] = [
|
|
{ text: "escape", toolCalls: [call("write_deliverable", { path: "../escape.txt", content: "x" })] },
|
|
{ text: "done", toolCalls: [] },
|
|
];
|
|
const result = await runAgentLoop(
|
|
{ providers: fakeProvider(script), tools: { db: connection.db, organization, chat: new ChatService(connection.db) } },
|
|
{ orgId, providerId, workspaceId: null, instructions: "i", initialUserTurn: "u", toolContext: ctx },
|
|
);
|
|
// The tool returns an error string to the model; nothing is written outside the root.
|
|
expect(existsSync(join(workspaceRoot, "..", "escape.txt"))).toBe(false);
|
|
expect(result.toolCallCount).toBe(1);
|
|
connection.close();
|
|
});
|
|
|
|
test("a first turn with no text and no tool calls fails the run instead of completing a no-op", async () => {
|
|
const connection = new DatabaseConnection(join(tmpdir(), `ac-loop3-${randomUUID()}.db`));
|
|
const organization = new OrganizationService(connection.db);
|
|
let orgId = "";
|
|
let ceoId = "";
|
|
let providerId = "";
|
|
connection.transaction(() => {
|
|
orgId = organization.createOrganization({ organizationName: "OrbitOps", employeeName: "Yang", mission: "m", goal: "g" });
|
|
providerId = (connection.db.prepare("select id from providers where org_id = ? and provider_key = 'openai'").get(orgId) as { id: string }).id;
|
|
ceoId = organization.createAgent({ orgId, name: "CEO", role: "CEO", department: "Executive", modelProviderId: providerId, modelId: "gpt-5.4", budgetCents: 25000, managerAgentId: null });
|
|
});
|
|
const ctx: AgentToolContext = {
|
|
orgId, agent: { id: ceoId, modelProviderId: providerId, modelId: "gpt-5.4" },
|
|
taskProjectId: null, taskGoalId: null, workspace: null,
|
|
};
|
|
await expect(runAgentLoop(
|
|
{ providers: fakeProvider([{ text: "", toolCalls: [] }]), tools: { db: connection.db, organization, chat: new ChatService(connection.db) } },
|
|
{ orgId, providerId, workspaceId: null, instructions: "i", initialUserTurn: "u", toolContext: ctx },
|
|
)).rejects.toThrow(/empty turn/);
|
|
connection.close();
|
|
});
|
|
});
|