nanocoai/nanoclaw / .claude/skills/add-cursor
add-cursor
FlaggedUse Cursor (local Agent SDK) as a full agent provider — planning, tool orchestration, MCP tools, session resume — alongside or instead of Claude. Cursor-account sign-in or API key, held by the credential gateway. Per-group via `ncl groups config update --provider cursor`. Distinct from using Cursor as a cloud agent (those run on Cursor VMs and are not this provider).
payload/container/agent-runner/src/providers/cursor.poll-loop.test.ts
View on GitHubimport { afterEach, beforeEach, describe, expect, it } from 'bun:test';
import fs from 'fs';
import path from 'path';
import { AgentNotFoundError } from '@cursor/sdk';
import { closeSessionDb, getInboundDb, initTestSessionDb } from '../mailbox/sqlite/connection.js';
import { getUndeliveredMessages } from '../db/messages-out.js';
import { getContinuation, setContinuation } from '../db/session-state.js';
import { GENERIC_FAILURE_NOTICE } from '../formatter.js';
import { MEMORY_SESSION_HOOK } from '../memory/session-hook.js';
import { processQuery, runPollLoop } from '../poll-loop.js';
import { CursorProvider, cursorAgentApi } from './cursor.js';
const TEST_ROOT = path.join(import.meta.dir, '..', '..', '.tmp-cursor-poll-loop');
const ROUTING = {
platformId: 'chan-1',
channelType: 'discord',
threadId: null,
inReplyTo: 'm1',
};
let previousHome: string | undefined;
let previousAgentDir: string | undefined;
beforeEach(() => {
fs.rmSync(TEST_ROOT, { recursive: true, force: true });
fs.mkdirSync(TEST_ROOT, { recursive: true });
previousHome = process.env.HOME;
previousAgentDir = process.env.NANOCLAW_AGENT_DIR;
process.env.HOME = fs.mkdtempSync(path.join(TEST_ROOT, 'home-'));
process.env.NANOCLAW_AGENT_DIR = fs.mkdtempSync(path.join(TEST_ROOT, 'agent-'));
initTestSessionDb();
getInboundDb()
.prepare(
`INSERT INTO destinations (name, display_name, type, channel_type, platform_id, agent_group_id)
VALUES ('discord-test', 'Discord Test', 'channel', 'discord', 'chan-1', NULL)`,
)
.run();
});
afterEach(() => {
if (previousHome === undefined) delete process.env.HOME;
else process.env.HOME = previousHome;
if (previousAgentDir === undefined) delete process.env.NANOCLAW_AGENT_DIR;
else process.env.NANOCLAW_AGENT_DIR = previousAgentDir;
closeSessionDb();
fs.rmSync(TEST_ROOT, { recursive: true, force: true });
});
function insertMessage(id: string, text: string): void {
getInboundDb()
.prepare(
`INSERT INTO messages_in (id, kind, timestamp, status, platform_id, channel_type, content)
VALUES (?, 'chat', datetime('now'), 'pending', 'chan-1', 'discord', ?)`,
)
.run(id, JSON.stringify({ sender: 'Alice', text }));
}
async function waitFor(condition: () => boolean, timeoutMs: number): Promise<void> {
const start = Date.now();
while (!condition()) {
if (Date.now() - start > timeoutMs) throw new Error('waitFor timeout');
await new Promise((resolve) => setTimeout(resolve, 50));
}
}
describe('CursorProvider through the real poll-loop', () => {
it('delivers assistant messages emitted before and after a tool call exactly once', async () => {
const original = { ...cursorAgentApi };
cursorAgentApi.create = async () =>
({
agentId: 'agent-stream',
send: async () => ({
supports: (op: string) => op === 'stream' || op === 'wait',
stream: async function* () {
yield {
type: 'assistant',
message: { content: [{ type: 'text', text: '<message to="discord-test">FIRST</message>' }] },
};
yield { type: 'tool_call', name: 'shell', status: 'running' };
yield {
type: 'assistant',
message: { content: [{ type: 'text', text: '<message to="discord-test">SECOND</message>' }] },
};
},
wait: async () => ({
id: 'run-stream',
status: 'finished' as const,
result: '<message to="discord-test">SECOND</message>',
}),
cancel: async () => {},
}),
close: () => {},
}) as never;
try {
const provider = new CursorProvider();
provider.registerMemorySessionHook(MEMORY_SESSION_HOOK);
const query = provider.query({ prompt: 'send two updates', cwd: process.env.NANOCLAW_AGENT_DIR! });
await processQuery(
query,
ROUTING,
['m1'],
'cursor',
undefined,
'send two updates',
undefined,
provider.emitsMidTurnText,
);
expect(getUndeliveredMessages().map((row) => JSON.parse(row.content).text)).toEqual(['FIRST', 'SECOND']);
} finally {
cursorAgentApi.create = original.create;
cursorAgentApi.resume = original.resume;
}
});
it('accepts the wrapping nudge and delivers the retried envelope', async () => {
const original = { ...cursorAgentApi };
const sent: string[] = [];
cursorAgentApi.create = async () =>
({
agentId: 'agent-wrap',
send: async (prompt: string) => {
sent.push(prompt);
return {
supports: (op: string) => op === 'wait',
wait: async () => ({
id: `run-${sent.length}`,
status: 'finished' as const,
result: sent.length === 1 ? 'bare text with no envelope' : '<message to="discord-test">WRAP_OK</message>',
}),
cancel: async () => {},
};
},
close: () => {},
}) as never;
try {
const provider = new CursorProvider();
provider.registerMemorySessionHook(MEMORY_SESSION_HOOK);
const query = provider.query({ prompt: 'say hi', cwd: process.env.NANOCLAW_AGENT_DIR! });
await processQuery(query, ROUTING, ['m1'], 'cursor', undefined, 'say hi', undefined);
expect(sent).toHaveLength(2);
expect(sent[1]).toContain('was not delivered');
const out = getUndeliveredMessages();
expect(out).toHaveLength(1);
expect(JSON.parse(out[0]!.content).text).toBe('WRAP_OK');
} finally {
cursorAgentApi.create = original.create;
cursorAgentApi.resume = original.resume;
}
});
it('delivers a terminal Cursor error without wrapping-nudging it', async () => {
const original = { ...cursorAgentApi };
const sent: string[] = [];
cursorAgentApi.create = async () =>
({
agentId: 'agent-error',
send: async (prompt: string) => {
sent.push(prompt);
return {
supports: (op: string) => op === 'wait',
wait: async () => ({
id: 'run-error',
status: 'error' as const,
error: { message: 'model failed mid-turn' },
}),
cancel: async () => {},
};
},
close: () => {},
}) as never;
try {
const provider = new CursorProvider();
provider.registerMemorySessionHook(MEMORY_SESSION_HOOK);
const query = provider.query({ prompt: 'fail', cwd: process.env.NANOCLAW_AGENT_DIR! });
await processQuery(query, ROUTING, ['m1'], 'cursor', undefined, 'fail', undefined);
expect(sent).toHaveLength(1);
const out = getUndeliveredMessages();
expect(out).toHaveLength(1);
expect(JSON.parse(out[0]!.content).text).toBe('model failed mid-turn');
} finally {
cursorAgentApi.create = original.create;
cursorAgentApi.resume = original.resume;
}
});
it('clears a stored Cursor continuation when resume throws AgentNotFoundError', async () => {
const original = { ...cursorAgentApi };
setContinuation('cursor', 'missing-agent');
insertMessage('m1', 'continue please');
cursorAgentApi.resume = async () => {
throw new AgentNotFoundError('missing');
};
try {
const provider = new CursorProvider();
provider.registerMemorySessionHook(MEMORY_SESSION_HOOK);
const controller = new AbortController();
const loopPromise = Promise.race([
runPollLoop({
provider,
providerName: 'cursor',
cwd: process.env.NANOCLAW_AGENT_DIR!,
signal: controller.signal,
}),
waitFor(() => getUndeliveredMessages().length > 0, 2000),
]);
await loopPromise;
controller.abort();
expect(getContinuation('cursor')).toBeUndefined();
const out = getUndeliveredMessages();
expect(out).toHaveLength(1);
expect(JSON.parse(out[0]!.content).text).toBe(GENERIC_FAILURE_NOTICE);
} finally {
cursorAgentApi.create = original.create;
cursorAgentApi.resume = original.resume;
}
});
});
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |