Files
orca/src/main/command-code/hook-service.test.ts
T
Brennan Benson 928d306b53 fix(agent-hooks): stop the Windows hook launcher spelling the AV-denied flag pair (STA-5237) (#16739)
* fix(agent-hooks): stop the Windows hook launcher spelling the AV-denied flag pair (STA-5237)

`-WindowStyle Hidden` + `-EncodedCommand` is denied at CreateProcess by
Kaspersky on Windows 11, whatever the payload decodes to. Bash reports it as
`Permission denied` and every managed hook event fails, so agent status never
arrives; the parent shell also briefly cannot spawn anything afterwards, so a
denied hook can take the user's next command down with it.

Measured on the reporting host (#16003), with a harmless `exit 0` payload:

  -NoProfile -ExecutionPolicy Bypass -WindowStyle Hidden -EncodedCommand  126
  -NoProfile -WindowStyle Hidden -EncodedCommand                          126
  -WindowStyle Hidden -EncodedCommand                                     126
  -NoProfile -EncodedCommand                                              0 (5/5)

#16576 removed `-ExecutionPolicy Bypass`, which is the one flag of the three
NOT in the signature, so hooks kept failing after that fix. The pair that has
to stop being spelled is `-WindowStyle Hidden` + `-EncodedCommand`.

Because the change is to the shared switch constant, it covers every site that
spells the denied pair in one edit: Claude via `wrapWindowsPowerShellEncodedCommand`,
gemini/cursor/droid/command-code/copilot via `wrapWindowsHookCommand`, the
`runtime-home-hook-command` unsafe-HOME fallback, and the spaced-path fallback
for codex/grok/devin/antigravity. Only a flag is removed, so parser and payload
compatibility is unchanged for every executor: the string is still a PowerShell
command line, still one self-contained token, still base64-shielded.

The tradeoff, recorded rather than hidden: `-WindowStyle Hidden` was the shipped
fix for #14815 (+#14828, #15117, #15447, #15767), and this removes it. Its
suppression was never measured — #14825 confirmed it visually, #16576's author
stated it "remains unverified on a real box", and #15506's author argued it
cannot help a `.cmd` child with no console to inherit. The console is allocated
by the parent chain, not by this command line. A live window measurement is
still outstanding and is called out in the PR.

Also adds `windows-hook-payload-delivery.test.ts` to the PR CI Windows leg,
which had never run it.

* test(agent-hooks): keep launcher token out of source grep
2026-08-27 10:48:53 -07:00

213 lines
7.6 KiB
TypeScript

import { afterEach, beforeEach, describe, expect, it, vi } from 'vitest'
import { spawn } from 'node:child_process'
import { mkdirSync, mkdtempSync, readFileSync, rmSync, writeFileSync } from 'node:fs'
import { createServer } from 'node:http'
import { tmpdir } from 'node:os'
import { dirname, join } from 'node:path'
import type { AddressInfo } from 'node:net'
const { homedirMock } = vi.hoisted(() => ({
homedirMock: vi.fn<() => string>()
}))
vi.mock('os', async () => {
const actual = (await vi.importActual('os')) as Record<string, unknown>
return {
...actual,
homedir: homedirMock
}
})
import { CommandCodeHookService } from './hook-service'
const WINDOWS_POWERSHELL_LAUNCHER =
/^[A-Za-z]:\/[^"]*\/System32\/WindowsPowerShell\/v1\.0\/powershell\.exe -NoProfile -EncodedCommand \S+$/
describe('CommandCodeHookService', () => {
let homeDir: string
beforeEach(() => {
homeDir = mkdtempSync(join(tmpdir(), 'orca-command-code-home-'))
homedirMock.mockReturnValue(homeDir)
})
afterEach(() => {
vi.clearAllMocks()
rmSync(homeDir, { recursive: true, force: true })
})
it('installs the managed command for Command Code status events', () => {
const status = new CommandCodeHookService().install()
expect(status.state).toBe('installed')
expect(status.managedHooksPresent).toBe(true)
const config = JSON.parse(
readFileSync(join(homeDir, '.commandcode', 'settings.json'), 'utf8')
) as {
hooks: Record<string, { matcher?: string; hooks: { command: string }[] }[]>
}
expect(Object.keys(config.hooks).sort()).toEqual(['PostToolUse', 'PreToolUse', 'Stop'].sort())
expect(config.hooks.PreToolUse[0].matcher).toBe('.*')
expect(config.hooks.PostToolUse[0].matcher).toBe('.*')
expect(config.hooks.Stop[0].matcher).toBeUndefined()
expect(config.hooks.PreToolUse[0].hooks[0].command).toMatch(
process.platform === 'win32' ? WINDOWS_POWERSHELL_LAUNCHER : /command-code-hook/
)
if (process.platform !== 'win32') {
expect(config.hooks.PreToolUse[0].hooks[0].command).toContain(join(homeDir, '.orca'))
}
if (process.platform !== 'win32') {
expect(config.hooks.PreToolUse[0].hooks[0].command).toMatch(/^if \[ -f /)
}
})
// Why: #6078 — a Windows user profile path with a space used to be written
// verbatim as the hook command, so the agent split it at the space. The
// managed command must use an encoded launcher so the path never appears raw
// on the cmd.exe command line.
it.skipIf(process.platform !== 'win32')(
'wraps the managed hook command to survive spaces in the profile path (#6078)',
() => {
const spaceHome = join(tmpdir(), 'orca command-code home with spaces')
mkdirSync(spaceHome, { recursive: true })
homedirMock.mockReturnValue(spaceHome)
try {
expect(new CommandCodeHookService().install().state).toBe('installed')
const config = JSON.parse(
readFileSync(join(spaceHome, '.commandcode', 'settings.json'), 'utf8')
) as { hooks: Record<string, { hooks: { command: string }[] }[]> }
const command = config.hooks.PreToolUse[0].hooks[0].command
expect(command).toMatch(WINDOWS_POWERSHELL_LAUNCHER)
} finally {
rmSync(spaceHome, { recursive: true, force: true })
}
}
)
it('installs a hook script that can recover the endpoint when Command Code strips token env', () => {
new CommandCodeHookService().install()
const scriptFileName =
process.platform === 'win32' ? 'command-code-hook.cmd' : 'command-code-hook.sh'
const script = readFileSync(join(homeDir, '.orca', 'agent-hooks', scriptFileName), 'utf8')
if (process.platform === 'win32') {
expect(script).toContain('sourceEndpointByPort')
expect(script).toContain('orca-dev\\agent-hooks')
expect(script).toContain('set ORCA_AGENT_HOOK_PORT=')
} else {
expect(script).toContain('Command Code strips TOKEN-like env vars')
expect(script).toContain('Command Code sanitizes hook subprocess env')
expect(script).toContain('__orca_read_ancestor_var')
expect(script).toContain('__orca_fill_from_endpoint_file')
expect(script).toContain('[ "$__orca_endpoint_port" != "$ORCA_AGENT_HOOK_PORT" ]')
expect(script).toContain('ORCA_PANE_KEY')
expect(script).toContain('ORCA_AGENT_LAUNCH_TOKEN')
expect(script).toContain('orca-dev/agent-hooks')
expect(script).toContain('endpoint_port=')
}
})
const itPosix = process.platform === 'win32' ? it.skip : it
itPosix('does not let stale endpoint files clobber recovered connection fields', async () => {
new CommandCodeHookService().install()
const staleEndpointPath = join(homeDir, 'stale-endpoint.env')
writeFileSync(
staleEndpointPath,
[
'ORCA_AGENT_HOOK_PORT=9',
'ORCA_AGENT_HOOK_TOKEN=stale-token',
'ORCA_AGENT_HOOK_ENV=development',
'ORCA_AGENT_HOOK_VERSION=1',
''
].join('\n')
)
const requests: { body: string; token: string | string[] | undefined }[] = []
const server = createServer((req, res) => {
let body = ''
req.setEncoding('utf8')
req.on('data', (chunk) => {
body += chunk
})
req.on('end', () => {
requests.push({
body,
token: req.headers['x-orca-agent-hook-token']
})
res.statusCode = 204
res.end()
})
})
try {
await new Promise<void>((resolve) => server.listen(0, '127.0.0.1', resolve))
const address = server.address() as AddressInfo
const scriptPath = join(homeDir, '.orca', 'agent-hooks', 'command-code-hook.sh')
const child = spawn('/bin/sh', [scriptPath], {
env: {
...process.env,
HOME: homeDir,
ORCA_AGENT_HOOK_ENDPOINT: staleEndpointPath,
ORCA_AGENT_HOOK_PORT: String(address.port),
ORCA_AGENT_HOOK_TOKEN: 'current-token',
ORCA_PANE_KEY: 'tab:leaf',
ORCA_TAB_ID: 'tab',
ORCA_WORKTREE_ID: 'worktree',
ORCA_AGENT_HOOK_ENV: 'development',
ORCA_AGENT_HOOK_VERSION: '1'
},
stdio: ['pipe', 'ignore', 'pipe']
})
child.stdin.end(JSON.stringify({ hook_event_name: 'Stop' }))
const exitCode = await new Promise<number | null>((resolve, reject) => {
let stderr = ''
child.stderr.setEncoding('utf8')
child.stderr.on('data', (chunk) => {
stderr += chunk
})
child.on('error', reject)
child.on('exit', (code) => {
if (stderr.trim()) {
reject(new Error(stderr))
} else {
resolve(code)
}
})
})
expect(exitCode).toBe(0)
expect(requests).toHaveLength(1)
expect(requests[0].token).toBe('current-token')
expect(new URLSearchParams(requests[0].body).get('paneKey')).toBe('tab:leaf')
} finally {
await new Promise<void>((resolve) => server.close(() => resolve()))
}
})
it('reports partial when one managed event is missing', () => {
const service = new CommandCodeHookService()
service.install()
const configPath = join(homeDir, '.commandcode', 'settings.json')
const config = JSON.parse(readFileSync(configPath, 'utf8')) as {
hooks: Record<string, unknown>
}
delete config.hooks.Stop
mkdirSync(dirname(configPath), { recursive: true })
writeFileSync(configPath, `${JSON.stringify(config, null, 2)}\n`)
const status = service.getStatus()
expect(status.state).toBe('partial')
expect(status.managedHooksPresent).toBe(true)
expect(status.detail).toBe('Managed hook missing for events: Stop')
})
})