Files
orca/src/shared/hermes-startup-query.ts
T
dcfa91919f fix(automations): deliver Hermes prompts via native query (#7862)
* fix(automations): deliver prompt to Hermes TUI via process-ready signal

Hermes's prompt_toolkit TUI never emits the DECSET 2004 bracketed-paste
handshake that the default draft-paste readiness waiter gates on, so the
automation prompt was silently dropped and the agent sat idle (terminal
opens, nothing entered). Add a 'process-ready' DraftPasteReadySignal that
arms the quiet-window on first PTY output and pastes once the TUI settles,
and assign it to the hermes agent config.

Adds a Hermes unit test covering the no-handshake path.

Design for the follow-up PR/issue review loop lives in
docs/design/pr-issue-review-loop.md (not committed; gitignored).

* fix(automations): skip broken process-name fallback for process-ready

CodeRabbit: the process-ready fallback consulted waitForAgentReady, which
compares the foreground process basename via isExpectedAgentProcess. Wrapped
interpreter launches surface as 'python3 .../hermes', so the check can never
confirm readiness and only drops the paste. process-ready readiness is the
PTY-quiet window (handled by waitForAgentDraftInputReady); the process-name
fallback is dead for this signal. Skip it and return false on timeout instead.

* test(automations): assert single paste for Hermes quiet-window path

Guard against a duplicate paste in the no-handshake Hermes path by
asserting sendRuntimePtyInputVerified is called exactly once. Addresses
the CodeRabbit nitpick on PR #7862.

* fix(automations): skip broken process-name fallback on pty-bound paste path

pasteDraftToAgentPtyWhenReady (quick-create/work-item route) still ran the
python3-vs-hermes process-name fallback that can never match for
process-ready agents, burning ~1s and dropping the paste. Mirror the
process-ready guard from pasteDraftWhenAgentReady and surface onTimeout.

Adds regression tests for both the quiet-window happy path and the
fallback-skip on the pty-bound path.

Co-authored-by: Orca <help@stably.ai>

* fix(automations): defer submit Enter until the Hermes TUI is interactive

Live-testing against Hermes v0.18.2 showed the fixed 50ms post-paste Enter
is swallowed: the node ui-tui takes 15s+ to boot, the paste fires ~1.5s in
(process-ready quiet window), and the cooked-mode line discipline turns the
early \r into \n, which the editor treats as newline-insert. The prompt
parked in the input box and the automation never executed.

For process-ready agents, defer the Enter until the TUI signals
interactivity (DECSET 2004 enable) or echoes the pasted content marker-free
(legacy prompt_toolkit), with a 120s best-effort cap. PTY input is FIFO, so
the deferred Enter always lands after the buffered paste text. The paste
echo of the content itself (raw or caret-notation markers adjacent) is
rejected so a cooked-mode echo can't release the Enter early.

Verified end-to-end in the dev app: automation prompt pasted, deferred
Enter released on tui-ready, Hermes submitted and ran the turn.

Co-authored-by: Orca <help@stably.ai>

* fix(automations): use Hermes native startup query

Co-authored-by: Orca <help@stably.ai>

* fix(automations): preserve quoted Hermes queries on Windows

* fix(agents): preserve invalid argument error message

Co-authored-by: Orca <help@stably.ai>

---------

Co-authored-by: Brandon Bennett <brandonbennett@macbookair.myfiosgateway.com>
Co-authored-by: Jinwoo-H <jinwoo0825@gmail.com>
Co-authored-by: Orca <help@stably.ai>
Co-authored-by: Jinwoo Hong <73622457+Jinwoo-H@users.noreply.github.com>
2026-07-11 02:35:01 -07:00

204 lines
7.1 KiB
TypeScript

import {
buildShellCommandFromArgv,
quoteStartupArg,
tokenizeStartupCommand,
type AgentStartupShell
} from './tui-agent-startup-shell'
const QUERY_ENV_LIMIT = 24_000
const QUERY_PLACEHOLDER = '__ORCA_HERMES_STARTUP_QUERY__'
const QUERY_ARG_PLACEHOLDER = `--query=${QUERY_PLACEHOLDER}`
const POSIX_QUERY_VARIABLE = '__orca_hermes_startup_query'
const POWERSHELL_QUERY_VARIABLE = 'orcaHermesStartupQuery'
const POWERSHELL_NATIVE_QUERY_VARIABLE = 'orcaHermesNativeQuery'
export const ORCA_HERMES_STARTUP_QUERY_ENV = 'ORCA_HERMES_STARTUP_QUERY'
function encodePowerShellCommand(command: string): string {
let bytes = ''
for (let index = 0; index < command.length; index += 1) {
const code = command.charCodeAt(index)
bytes += String.fromCharCode(code & 0xff, code >>> 8)
}
return btoa(bytes)
}
function encodePosixEvalScript(command: string): string {
return Array.from(
new TextEncoder().encode(command),
(byte) => `\\0${byte.toString(8).padStart(3, '0')}`
).join('')
}
function tokenizeCommand(value: string, shell: AgentStartupShell): string[] | null {
const tokenized = tokenizeStartupCommand(value, shell)
return tokenized.ok && tokenized.tokens.length > 0 ? tokenized.tokens : null
}
function isHermesExecutableToken(value: string): boolean {
const basename = value.replaceAll('\\', '/').split('/').pop()?.toLowerCase() ?? ''
return basename === 'hermes' || basename === 'hermes.exe' || basename === 'hermes.cmd'
}
const HERMES_VALUELESS_FLAGS = new Set([
'--tui',
'--cli',
'--verbose',
'-v',
'--quiet',
'-Q',
'--worktree',
'-w',
'--accept-hooks',
'--checkpoints',
'--yolo',
'--pass-session-id',
'--ignore-user-config',
'--ignore-rules',
'--safe-mode',
'--dev'
])
function findChatSubcommand(args: readonly string[]): number {
for (let index = 0; index < args.length; index += 1) {
const token = args[index]
if (token === 'chat') {
return index
}
if (token.startsWith('-') && !token.includes('=') && !HERMES_VALUELESS_FLAGS.has(token)) {
index += 1
}
}
return -1
}
function stripOrcaOwnedHermesArgs(args: readonly string[]): string[] {
const normalized: string[] = []
for (let index = 0; index < args.length; index += 1) {
const token = args[index]
if (token === '--query' || token === '-q') {
index += 1
continue
}
if (
token.startsWith('--query=') ||
token.startsWith('-q=') ||
(token.startsWith('-q') && token.length > 2) ||
token === '--tui' ||
token === '--cli'
) {
continue
}
normalized.push(token)
}
return normalized
}
function normalizeHermesArgv(
baseArgv: string[],
configuredArgv: string[],
shell: AgentStartupShell
): string[] | null {
const executableCandidates: number[] = []
for (let index = 0; index < baseArgv.length; index += 1) {
if (isHermesExecutableToken(baseArgv[index])) {
executableCandidates.push(index)
}
}
let anchoredExecutable: number | undefined
for (const index of executableCandidates) {
if (findChatSubcommand(baseArgv.slice(index + 1)) !== -1) {
// Wrapper arguments can also be named "hermes"; the final match is the executable.
anchoredExecutable = index
}
}
const executableIndex = anchoredExecutable ?? executableCandidates.at(-1) ?? -1
if (executableIndex === -1) {
return null
}
let commandPrefix = baseArgv.slice(0, executableIndex + 1)
let assignmentCount = 0
while (/^[A-Za-z_][A-Za-z0-9_]*=/.test(commandPrefix[assignmentCount] ?? '')) {
assignmentCount += 1
}
if (assignmentCount > 0) {
if (shell !== 'posix') {
return null
}
commandPrefix = ['env', ...commandPrefix]
}
const baseArgs = baseArgv.slice(executableIndex + 1)
const chatIndex = findChatSubcommand(baseArgs)
const baseArgsWithoutChat =
chatIndex === -1 ? baseArgs : baseArgs.filter((_, index) => index !== chatIndex)
const configuredChatIndex = findChatSubcommand(configuredArgv)
const configuredArgsWithoutChat =
configuredChatIndex === -1
? configuredArgv
: configuredArgv.filter((_, index) => index !== configuredChatIndex)
return [
...commandPrefix,
'chat',
QUERY_ARG_PLACEHOLDER,
...stripOrcaOwnedHermesArgs(baseArgsWithoutChat),
...stripOrcaOwnedHermesArgs(configuredArgsWithoutChat),
'--tui'
]
}
function buildQueryCommand(argv: string[], shell: AgentStartupShell): string {
if (shell !== 'posix') {
const invocation = buildShellCommandFromArgv(argv, 'powershell').replace(
quoteStartupArg(QUERY_ARG_PLACEHOLDER, 'powershell'),
`"--query=$${POWERSHELL_NATIVE_QUERY_VARIABLE}"`
)
// Why: startup prompts must not remain exported to Hermes tools; the
// long-lived parent shell retains the transport env for compatibility.
// PowerShell 5 needs Windows-native quote escaping before building child argv.
const script = `$${POWERSHELL_QUERY_VARIABLE} = $env:${ORCA_HERMES_STARTUP_QUERY_ENV}; $${POWERSHELL_NATIVE_QUERY_VARIABLE} = $${POWERSHELL_QUERY_VARIABLE} -replace '(\\\\*)"', '$1$1\\"'; Remove-Item Env:${ORCA_HERMES_STARTUP_QUERY_ENV} -ErrorAction SilentlyContinue; ${invocation}`
return `powershell.exe -NoProfile -EncodedCommand ${encodePowerShellCommand(script)}`
}
const invocation = buildShellCommandFromArgv(argv, 'posix').replace(
quoteStartupArg(QUERY_ARG_PLACEHOLDER, 'posix'),
`"--query=\${${POSIX_QUERY_VARIABLE}}"`
)
const encodedInvocation = encodePosixEvalScript(invocation)
// Why: a fixed single-quote-safe wrapper parses in POSIX shells and pwsh;
// the dynamic argv is decoded only after entering the known `sh` grammar.
const script = `${POSIX_QUERY_VARIABLE}="\${${ORCA_HERMES_STARTUP_QUERY_ENV}}"; unset ${ORCA_HERMES_STARTUP_QUERY_ENV}; eval "$(printf %b "${encodedInvocation}")"`
return `sh -c ${quoteStartupArg(script, 'posix')}`
}
export function planHermesStartupQuery(args: {
baseCommand: string
agentArgs?: string | null
prompt: string
agentEnv?: Record<string, string> | null
platform: NodeJS.Platform
shell: AgentStartupShell
isRemote?: boolean
}): { command: string; env: Record<string, string> } | null {
const baseArgv = tokenizeCommand(args.baseCommand, args.shell)
const configuredArgv = args.agentArgs?.trim() ? tokenizeCommand(args.agentArgs, args.shell) : []
if (!baseArgv || !configuredArgv) {
return null
}
const argv = normalizeHermesArgv(baseArgv, configuredArgv, args.shell)
if (!argv) {
return null
}
const command = buildQueryCommand(argv, args.shell)
const env = { ...args.agentEnv, [ORCA_HERMES_STARTUP_QUERY_ENV]: args.prompt }
const envSize = Object.entries(env).reduce((total, [key, value]) => {
if (args.platform === 'win32') {
return total + key.length + value.length + 2
}
return total + new TextEncoder().encode(`${key}=${value}`).byteLength + 1
}, 0)
const commandSize =
args.platform === 'win32' ? command.length : new TextEncoder().encode(command).byteLength
// Why: WSL plans execute as Linux but cross the Windows environment block;
// the conservative shared bound is safe for every transport host.
return commandSize + envSize <= QUERY_ENV_LIMIT ? { command, env } : null
}