Files
orca/src/main/codex/codex-app-server-session.ts
T
Brennan Benson fd9125ea8c feat(native-chat): Codex structured native chat restructure (#16729)
* feat(native-chat): port structured Codex sessions from restructure-recovery

Rebuilds the desktop structured native-chat implementation from
brennanb2025/native-chat-restructure-recovery (tip 4e31c08db3) on top of
current main as a single commit, scoped to the local Codex path.

Ported:
- Structured agent-session core: durable record store + single-writer lease,
  canonical journal, agent-session wire host/attach/eviction/subscribers,
  `agentSession.*` RPC surface (registered via ALL_RPC_METHODS; host-side
  mobile allowlist included for wire compat), pty write gate, transcript
  additions, and the Codex app-server adapter/launch resolution.
- Renderer: NativeChatStructuredSession view/composer stack, structured
  launch path with the single-flight guard, local structured session tabs
  sync, activation gate + structured inventory (read-only
  `agentSession.handoffStatus` probe), agent-session tabs in the tab strip,
  AI-vault structured session activation, and the settings pane with the
  parent Experimental Chat UI toggle plus the nested "Use updated structured
  native chat" toggle. New sessions require both flags, agent codex, no
  prompt, and a local non-WSL, non-Windows-host execution host
  (structured-native-chat-availability).
- Fixes 72c013cea6 (verified Codex launch recovery), 8ddbaf5e3d (defer
  native terminal view switching affordances), and 4e31c08db3 (release the
  launch gate after a visibility retry) with their regression tests,
  including the third-launch-after-retry guard case.
- Cross-version agent-session wire test + CI lane, packaging entries
  (proper-lockfile, agent-tooling asar excludes), and the wire-compat doc
  section.

Deliberately not ported: mobile/ changes, the Claude structured runtime
(only the claude-transcript-branch-proof and claude-structured-owner-identity
leaf modules remain, backing the kept TUI-recovery arms), the terminal↔chat
adoption/handoff flow (`agentSession.adoptTerminal`/`requestHandoff`, the
handoff request engine, TUI adoption machinery, orca-runtime adoption
methods), renderer switching affordances and their dead leftovers, the
hook/subagent-status refactor cluster, and unrelated branch changes. The
crash-during-acquisition recovery path (restart handoff adjudication,
restore/reverse re-acquire, lease schema handoff keys) is kept because every
plain direct launch depends on it; a trimmed handoff coordinator exposes
only status/restore/close.

Branch edits that targeted files main has since split (ipc/pty.ts,
worktrees.ts, rpc/methods/terminal.ts, useIpcEvents, pty-connection,
store/slices/terminals.ts, runtime-types, web preload) were re-applied to
the split modules, preserving main's newer logic (Windows CIM fallback,
browser tab close rework, cold-restore resume flow, dispatcher threading).

Known seam: the mobile clipboard image-provenance CONSUMER gate ships
(agentSession.send refuses unproven mobile image refs with
agent_session_image_untrusted) but the producer hunk in
rpc/methods/clipboard.ts stays with the unported mobile cluster, so mobile
image sends into structured chat fail closed until that side ports.

* fix(native-chat): trust only authenticated local image uploads

* fix(build): preserve Windows process-tree patch application

* test(windows): include process creation time in addon fixture

* fix(build): run windows-process-tree node-gyp from the physical package dir

gyp expands the node-addon-api dependency by probing node, whose cwd
resolves to the package's physical directory in the store, so the emitted
target is a store-relative ../../../../node-addon-api@... hop. gyp then
resolves that hop against the rebuild cwd; from the node_modules
symlink/junction it escapes the store and configure fails with
"node_addon_api.gyp not found" (run 32999886072).

Rebuild from realpath(package dir) so both bases agree, matching how the
package manager itself runs native install scripts. The regression test
replays gyp's expansion+resolution against the planned cwd and fails
without the fix.

* fix(native-chat): keep chat tabs visible through terminal closes and empty-worktree launches

Two proven blockers in the native Codex tab contract:

closeTerminalTab pre-empted the canonical unified close. With one terminal
left it deactivated the worktree on a terminal/editor/browser-only check,
blanking a workspace that still held a renderable agent-session tab; with
two or more it pre-picked a successor from terminal entities only,
re-stamping the group active before closeUnifiedTab's MRU/neighbor repair
could land on the chat tab. Successor choice now defers to the unified
contract whenever the terminal has a unified row, and deactivation is
gated on the unified renderable count (matching leaveWorktreeIfEmpty),
with the legacy pre-pick kept only for terminals without a unified row.

A structured session created on an empty worktree was published into the
host's headless group while preserveLocalLayout froze the local layout,
leaving the tab in store but permanently off screen. A preserveLocalLayout
owner now always takes client-owned placement — repairing a rendered
leaf whose group record is missing, or materializing a rendered group on a
truly empty worktree — and applies the client-derived layout repair while
still rejecting host-authored layout.

Regression tests drive the real store through closeTerminalTab (git
worktree and folder workspace) and the real snapshot applier for the
empty-worktree adoption states; all fail without the fixes.

* fix(native-chat): close stale turns and retry rejected sends

* fix(native-chat): retire hosted rows on structured tab activation

* fix(native-chat): preserve rpc defaults across main merge

* chore: format remote wire compatibility guide

* test(native-chat): cover retry after unconfirmed send

* fix(native-chat): reload outbox on session switch

* docs(settings): disclose structured chat platform limits

* fix(native-chat): await Codex launch-home preparation

* fix(codex): align child-process allowlist with async trust bridge

* test(identity): update inventory for tab surface refactor

* fix(windows): preserve process-tree CRLF patch sources

* fix(native-chat): anchor an unmatched chat echo where it was sent (#16117)

* fix(native-chat): anchor an unmatched chat echo where it was sent

The reported symptom was old user messages replaying below every new turn, so the
conversation read as scrambled. The cause was not that the echo failed to match a
transcript row. Claude consumes a mid-turn send through a `queued_command`
attachment and writes no `type:"user"` record for it, so some echoes can never
match, and no amount of matching will change that. The cause was WHERE an
unmatched echo rendered: buildMobileNativeChatTransientData appended every pending
item after the entire transcript, so it re-read below each turn that landed
afterwards.

Render each echo directly after the transcript row it was sent against, using the
baseline the send already captures. An unmatched echo is then at worst a duplicate
in the right position rather than a scrambled one, and it stays visible. Echoes
sharing an anchor keep send order; a send with no baseline, or one whose anchor
folding dropped, still falls back to the tail.

Deliberately NOT fixed by deleting the echo. Inferring from send ordering that an
echo can never match, then removing it, loses the user's own text for a message
the agent did receive, and it cannot fire in the common case anyway - measured
drain groups are 1,017 of size 1 against 55 larger. It also escalates an existing
gap: the count pass has no baseline-tail guard, unlike the glue pass, while
`messages` is a 40-row window that head-trims, resets on reconnect and grows at
the front on loadEarlier, so a false landing there would license deleting a
DIFFERENT outstanding message.

That count-pass gap is real and left for a separate change; anchoring makes its
worst case a duplicate in place rather than a scrambled conversation.

* fix(native-chat): preserve folded echo anchors

* fix(native-chat): preserve forward-folded echo anchors

* fix(native-chat): keep leading folded echoes in place

* fix(workspace-cleanup): show git status for every row (#16690)

* fix(native-chat): refuse structured chat on every Windows execution path

canUseStructuredNativeChat only refused win32 when a project runtime
resolved, so folder-workspace keys (and other keys with no project
runtime) failed open into structured chat on Windows. Fail closed on
win32 unconditionally after the host check, matching the settings copy:
local macOS/Linux only; Windows/WSL/SSH stay on terminal chat.

* fix(native-chat): restore runtime refusals behind the win32 gate

506d375de3 replaced the project-runtime checks with a bare platform test,
so a WSL or repair-required runtime resolution would no longer refuse
structured chat off-win32. Keep the unconditional win32 refusal and
re-run the runtime resolution after it, so the gate does not depend on
the resolver's own platform guard. Tests inject WSL and repair-required
resolutions on darwin/linux and fail against the regressed gate.

* fix structured session journal durability

* fix structured tab active pointer after restart

* fix(native-chat): await optional lease renewal callbacks

* refactor(skills): extract install error messages

* fix(agent-session): harden recovery ownership

* fix(native-chat): retain panes across tab activation

* fix(native-chat): address round-one review findings

* test(native-chat): align integration coverage after main merge

* fix(native-chat): harden round-two reliability

* fix(native-chat): harden round-three reliability

* fix(native-chat): close round-four recovery gaps

* fix(native-chat): separate bounded journal key forms

* fix(native-chat): reset outbox error in render on session switch

The switch effect adjusted error state after the sessionId prop changed,
tripping react-doctor's no-adjust-state-on-prop-change on the changed-code
gate and flashing the old session's banner for a frame. Reset it with the
render-time previous-value guard instead.

* fix(native-chat): invalidate stale outbox settlements

* test(native-chat): restore settled-error session-switch regression

a6e2379bd1 replaced this test with the in-flight settlement race test,
leaving the render-time error reset unpinned: deleting the reset block
still passed the whole native-chat suite. Keep both scenarios pinned;
they are distinct (settled error clears on switch vs stale settlement
invalidated in the commit-to-passive window).

* test(wire): make release checkouts race safe

* test(wire): pin cross-process checkout single-flight and importer specifier contract

* test(wire): harden release checkout lifecycle

* fix(build): drop CR-byte residue from windows-process-tree patch

The two trailing CR bytes on the patch's deletion lines are a proven
no-op: pnpm hashes patches CRLF-normalized (both forms hash to the
lockfile's 946ffb2b) and materializes this package without applying the
patch in either form, so the load-bearing build edits come solely from
applyWindowsProcessTreeBuildFixes() (#16947), which handles both source
EOL forms. Restore byte-identity with main and repin the contract test
to the post-#16947 reality: LF-only patch bytes plus lockfile hash sync.

* fix(native-chat): skip empty startup recovery
2026-08-28 16:45:58 -07:00

356 lines
12 KiB
TypeScript

import { spawn, type ChildProcess, type ChildProcessWithoutNullStreams } from 'node:child_process'
import { waitForProcessExitUntil } from './codex-process-exit-deadline'
import { stderrIndicatesMissingAppServer } from './codex-app-server-capability-signal'
import { withCliRuntimeOnPath } from '../../shared/node-cli-command-resolution'
// Why: `codex app-server` is Orca's sanctioned RPC surface into Codex-owned
// state (hook trust hashes, the sqlite thread index). This module owns the
// stdio JSONL transport — spawn, handshake, framing, deadline, reap — so every
// RPC consumer (trust grant, session index heal) shares one hardened lifecycle.
export type CodexAppServerInvocation = {
command: string
args: string[]
/**
* The resolved CLI path, used to pair the CLI with the `node` it was installed
* against — without it a CLI resolved out of a version-manager directory runs
* under whatever node leads PATH and dies on a NODE_MODULE_VERSION mismatch
* (stablyai/orca#10932).
*
* Required, and `null` only for a guest-side launcher (wsl.exe) where the host
* path means nothing. Optional would let a native builder omit it and silently
* fall back to pairing against a cmd.exe wrapper with no type error.
*/
cliPath: string | null
/** Overlay applied on top of the inherited environment (e.g. CODEX_HOME). */
env?: Record<string, string>
/** Env keys stripped from the inherited environment before spawn (e.g. an
* inherited CODEX_HOME, so a default-home grant runs against the real ~/.codex). */
envToDelete?: readonly string[]
/** Whole-session deadline. The codex child is SIGKILLed when it lapses. */
timeoutMs: number
}
/** Codex-side absence of the requested app-server RPC surface (old CLI without
* the app-server subcommand, or a server without the called methods).
* This is the ONLY error class capability caches mark unsupported. */
export class CodexAppServerUnsupportedError extends Error {
constructor(message: string) {
super(message)
this.name = 'CodexAppServerUnsupportedError'
}
}
export class CodexAppServerTimeoutError extends Error {
constructor(message: string) {
super(message)
this.name = 'CodexAppServerTimeoutError'
}
}
export function isCodexAppServerUnsupportedError(error: unknown): boolean {
return error instanceof Error && error.name === 'CodexAppServerUnsupportedError'
}
type JsonRpcResponse = {
id?: number
result?: unknown
error?: { code?: number; message?: string }
}
export type CodexAppServerRpc = {
request: (method: string, params?: Record<string, unknown>) => Promise<unknown>
notify: (method: string, params?: Record<string, unknown>) => void
}
const JSON_RPC_METHOD_NOT_FOUND = -32601
const STDERR_TAIL_MAX_BYTES = 8192
const STDOUT_LINE_MAX_BYTES = 1024 * 1024
export function killCodexAppServerProcessTree(
child: Pick<ChildProcess, 'pid' | 'kill'>,
options: { platform?: NodeJS.Platform; spawnImpl?: typeof spawn } = {}
): void {
const platform = options.platform ?? process.platform
const spawnImpl = options.spawnImpl ?? spawn
if (platform === 'win32' && child.pid) {
try {
// Why: npm-installed Codex runs behind cmd.exe; killing only that wrapper
// leaves the app-server child alive after a timeout or failed shutdown.
const killer = spawnImpl('taskkill', ['/pid', String(child.pid), '/t', '/f'], {
stdio: 'ignore',
windowsHide: true
})
let fellBack = false
const killDirectChild = (): void => {
if (!fellBack) {
fellBack = true
child.kill('SIGKILL')
}
}
killer.on('error', killDirectChild)
killer.on('exit', (code) => {
if (code !== 0) {
killDirectChild()
}
})
killer.unref()
return
} catch {
// Fall through to the direct-child best effort when taskkill cannot start.
}
}
if (child.pid) {
try {
// npm/package-manager launchers insert a shim child on POSIX. Reap its
// direct descendants before signalling the wrapper itself.
const descendants = spawnImpl('pkill', ['-KILL', '-P', String(child.pid)], {
stdio: 'ignore'
})
// A missing pkill surfaces as an async 'error' event, and an unhandled one
// takes down the main process.
descendants.on('error', () => undefined)
descendants.unref()
} catch {
// The direct kill below remains the fallback when pkill is unavailable.
}
}
child.kill('SIGKILL')
}
/** Codex answering "no such method" is the only response that proves the RPC
* surface is absent rather than temporarily failing. */
export function isCodexMethodNotFoundError(error: unknown): boolean {
if (typeof error !== 'object' || error === null) {
return false
}
const { code, message } = error as { code?: unknown; message?: unknown }
return (
code === JSON_RPC_METHOD_NOT_FOUND ||
/method not found/i.test(typeof message === 'string' ? message : '')
)
}
/**
* Runs one short-lived `codex app-server` session over stdio JSON-RPC (JSONL):
* spawn → initialize → initialized → body(rpc) → EOF/reap. The child is reaped
* on every path; the session deadline SIGKILLs it.
*/
export async function runCodexAppServerSession<T>(
invocation: CodexAppServerInvocation,
body: (rpc: CodexAppServerRpc) => Promise<T>,
spawnImpl: typeof spawn = spawn
): Promise<T> {
// Why: a default-home grant must run against the real ~/.codex, so strip an
// inherited CODEX_HOME (envToDelete) after applying the overlay, not before.
const childEnv: NodeJS.ProcessEnv = { ...process.env, ...invocation.env }
for (const key of invocation.envToDelete ?? []) {
delete childEnv[key]
}
const pairedEnv = invocation.cliPath
? withCliRuntimeOnPath(invocation.cliPath, childEnv)
: childEnv
const child = spawnImpl(invocation.command, invocation.args, {
env: pairedEnv,
stdio: ['pipe', 'pipe', 'pipe'],
windowsHide: true
}) as ChildProcessWithoutNullStreams
let stderrTail = ''
let exited = false
let nextRequestId = 1
let timedOut = false
const pending = new Map<
number,
{ resolve: (r: JsonRpcResponse) => void; reject: (e: Error) => void }
>()
const exitPromise = new Promise<void>((resolve) => {
child.on('exit', () => {
exited = true
resolve()
})
})
// Why: 'error' fires instead of 'exit' when the spawn itself fails
// (ENOENT); surface it to every in-flight request or they wait forever.
let spawnError: Error | null = null
child.on('error', (error) => {
spawnError = error
exited = true
failPending(error)
})
// Why: 'close' (not 'exit') guarantees the stderr tail is complete, so an
// early death classifies correctly as missing-subcommand vs transient.
child.on('close', () => {
failPending(buildEarlyExitError())
})
// Why: JSONL can contain non-ASCII hook paths. Stream decoding must retain a
// multibyte character split across pipe chunks or the response becomes invalid JSON.
child.stderr.setEncoding('utf8').on('data', (chunk: string) => {
stderrTail = (stderrTail + chunk).slice(-STDERR_TAIL_MAX_BYTES)
})
// Why: a child can exit between the liveness check and stdin.write(); an
// EPIPE must reject the RPC instead of becoming an unhandled stream error.
child.stdin.on('error', (error) => {
failPending(error)
})
let stdoutBuffer = ''
child.stdout.setEncoding('utf8').on('data', (chunk: string) => {
stdoutBuffer += chunk
if (Buffer.byteLength(stdoutBuffer) > STDOUT_LINE_MAX_BYTES) {
// Why: Windows process-tree termination is asynchronous; stop buffered
// chunks from spawning another taskkill for the same oversized response.
child.stdout.destroy()
killCodexAppServerProcessTree(child)
failPending(new Error('codex app-server emitted an oversized JSONL response'))
return
}
let newlineIndex
while ((newlineIndex = stdoutBuffer.indexOf('\n')) !== -1) {
const line = stdoutBuffer.slice(0, newlineIndex).trim()
stdoutBuffer = stdoutBuffer.slice(newlineIndex + 1)
if (!line) {
continue
}
let message: JsonRpcResponse
try {
message = JSON.parse(line) as JsonRpcResponse
} catch {
continue
}
if (typeof message.id === 'number' && pending.has(message.id)) {
const waiter = pending.get(message.id)!
pending.delete(message.id)
waiter.resolve(message)
}
}
})
function failPending(error: Error): void {
for (const waiter of pending.values()) {
waiter.reject(error)
}
pending.clear()
}
let rejectDeadline: (error: Error) => void = () => {}
const deadlinePromise = new Promise<never>((_resolve, reject) => {
rejectDeadline = reject
})
const deadline = setTimeout(() => {
timedOut = true
const error = new CodexAppServerTimeoutError(
`codex app-server session exceeded ${invocation.timeoutMs}ms (${invocation.command})`
)
killCodexAppServerProcessTree(child)
failPending(error)
rejectDeadline(error)
}, invocation.timeoutMs)
function sendLine(payload: Record<string, unknown>): void {
child.stdin.write(`${JSON.stringify(payload)}\n`)
}
function notify(method: string, params?: Record<string, unknown>): void {
const payload: Record<string, unknown> = { method }
if (params !== undefined) {
payload.params = params
}
try {
sendLine(payload)
} catch {
// Notifications are fire-and-forget; a dead child fails the next request.
}
}
async function requestRpc(method: string, params?: Record<string, unknown>): Promise<unknown> {
if (spawnError) {
throw spawnError
}
if (timedOut) {
throw new CodexAppServerTimeoutError('codex app-server session already timed out')
}
if (exited) {
throw buildEarlyExitError()
}
const id = nextRequestId++
const response = await new Promise<JsonRpcResponse>((resolve, reject) => {
pending.set(id, { resolve, reject })
const payload: Record<string, unknown> = { method, id }
if (params !== undefined) {
payload.params = params
}
try {
sendLine(payload)
} catch (error) {
pending.delete(id)
reject(error instanceof Error ? error : new Error(String(error)))
}
})
if (response.error) {
if (isCodexMethodNotFoundError(response.error)) {
throw new CodexAppServerUnsupportedError(
`codex app-server does not support ${method}: ${response.error.message ?? 'method not found'}`
)
}
throw new Error(
`codex app-server ${method} failed: ${response.error.message ?? 'unknown error'}`
)
}
return response.result
}
function buildEarlyExitError(): Error {
if (stderrIndicatesMissingAppServer(stderrTail)) {
return new CodexAppServerUnsupportedError(
`codex CLI does not support the app-server subcommand: ${stderrTail.trim().slice(0, 400)}`
)
}
return new Error(
`codex app-server exited before completing the session${stderrTail ? `: ${stderrTail.trim().slice(0, 400)}` : ''}`
)
}
try {
const session = async (): Promise<T> => {
await requestRpc('initialize', {
clientInfo: { name: 'orca_desktop', title: 'Orca', version: '0.0.0' }
})
notify('initialized')
return body({ request: requestRpc, notify })
}
// Why: the timeout owns the whole callback, including time between RPCs;
// killing the child alone cannot settle a callback awaiting unrelated work.
return await Promise.race([session(), deadlinePromise])
} catch (error) {
if (
error instanceof Error &&
!(error instanceof CodexAppServerUnsupportedError) &&
!(error instanceof CodexAppServerTimeoutError) &&
stderrIndicatesMissingAppServer(stderrTail)
) {
throw new CodexAppServerUnsupportedError(
`codex CLI does not support the app-server subcommand: ${stderrTail.trim().slice(0, 400)}`
)
}
throw error
} finally {
try {
child.stdin.end()
} catch {
// stdin may already be destroyed after a kill; reaping below still runs.
}
if (!exited) {
// Why: the server exits promptly on stdin EOF; the grace period only
// bounds a wedged child before the guaranteed SIGKILL reap.
await waitForProcessExitUntil(exitPromise, 1500)
if (!exited) {
killCodexAppServerProcessTree(child)
await waitForProcessExitUntil(exitPromise, 1000)
}
}
clearTimeout(deadline)
}
}