fix(antigravity): heal empty-projectId accounts via retryable auto-onboarding (#10424)

* fix(antigravity): heal empty-projectId accounts via retryable auto-onboarding

Accounts with an empty Cloud Code projectId get a permanent 422 "Missing
Google projectId" when loadCodeAssist returns no project. The 3.8.50
bootstrap attempts to CREATE the project via onboardUser, but a single failed
attempt (transient network/upstream error) was memoized forever in
onboardAttemptedCache: every later request in the process skipped onboarding
and 422'd, even though a retry would succeed.

Replace the permanent per-token Set with a failure-backoff map: failed onboard
attempts are retried after a 5-minute backoff (bounded, self-healing), the
in-flight lock still dedupes concurrent calls, and success clears the failure
marker and memoizes the project as before. Accounts that CAN be onboarded now
heal automatically on a later request or token refresh — no user action.

Tests: the existing "does not retry" case is now framed as the backoff window;
a new case proves the account heals (retries onboarding and recovers the
project) once the backoff expires.

* chore(changelog): fragment for #10424 antigravity project autocreate

* feat(antigravity): BYOP fast-fail + manual GCP project-id override

Port decolua/9router#2934 + VansRouter 802a859:
- tryOnboardUser now returns a three-way status; a 200 onboardUser response
  WITHOUT cloudaicompanionProject means Google deprecated automatic project
  creation for standard-tier (personal) accounts (BYOP). Such accounts are
  cached permanently (no pointless ~18s re-onboard) and the executor fails
  fast with 403 GCP_PROJECT_REQUIRED + actionable 'enter your project id'
  message instead of the generic 422 or a delayed 429.
- Transient onboard failures keep the existing 5-min backoff heal.
- Manual project-id override: the EditConnectionModal now stamps
  providerSpecificData.isProjectIdManual when the operator enters a project
  id, and tokenRefresh skips auto-discovery for flagged accounts so the
  manual value is never overwritten.

* chore(changelog): cover BYOP fast-fail + manual override in #10424 fragment

* test(antigravity): expect fast 403 GCP_PROJECT_REQUIRED when loadCodeAssist finds no project (#10424)

Google now marks accounts without an onboarded project as BYOP (automatic
project creation deprecated for standard-tier accounts, #2934). The PR's
BYOP fast-fail path returns 403 gcp_project_required instead of the old
generic 422 missing_project_id; align the #2334 executor test with that
contract so CI unit-test shard 2/4 passes.

* fix(antigravity): persist isProjectIdManual, fix BYOP citation, dodge refresh-retry

Review follow-up on #10424:

1. EditConnectionModal: isProjectIdManual was set on
   updates.providerSpecificData right after the project-id field, then the
   OAuth path (Antigravity is always OAuth) rebuilt providerSpecificData from
   connection.providerSpecificData before the request went out, discarding the
   flag — tokenRefresh.ts was guarding a field never actually persisted. The
   flag now lands in the single surviving antigravity merge, with a jsdom
   regression test (modeled on edit-connection-modal-openai-store-toggle).

2. The '#2934' citation for the Google BYOP claim pointed at an unrelated
   closed issue. Swapped for the real tracking issue #8491 (empty Google
   projectId -> 422 class) across bootstrap/executor/test comments.

3. BYOP fast-fail now returns 422 instead of 403: chatCore's generic
   401/403 -> refresh-and-retry path was hitting Google's OAuth token
   endpoint on every request from an affected account (pointless — refreshing
   cannot create a GCP project), and 422 matches the sibling
   missing_project_id error the client already maps to an action-needed
   prompt.

Also: eslint-disable-next-line for the pre-existing
react-hooks/set-state-in-effect baseline noise in the modal (repo
convention, same pattern as 11 other dashboard files).

* chore(ci): drop unused eslint-disable in EditConnectionModal form hydration

The react-hooks/set-state-in-effect disable added in the previous commit is
unused under the repo's pinned eslint-plugin-react-hooks (7.0.1) — the rule
does not fire on this line at that version, so the unused directive tripped
the whole-repo 'No new ESLint warnings' gate (max-warnings 0). Verified with
the lockfile-pinned plugin: lint:json is clean (0 errors, 0 warnings).

* fix(build): bound and retry the opencode-plugin npm install in prepublish

The plugin's node_modules is gitignored, so every fresh CI checkout runs a
full npm install inside @omniroute/opencode-plugin during build:cli. npm's
unbounded fetch retries turn a stalled registry CDN connection (the recurring
onnxruntime-class ETIMEDOUT flake) into a 20-30 minute hang — the DAST
'Build CLI bundle' step has been cancelled at the 30m cap repeatedly.

- Bound npm fetch: --fetch-timeout 60s, 2 retries with capped backoff — a
  stalled connection now fails fast instead of hanging the job.
- Retry the install up to 3 times with a 10s pause between attempts, so
  transient CDN failures recover in-build.

Net effect: the step either completes (network OK) or fails quickly with a
clear error (network down) — it can no longer eat the whole job budget.

* ci(quality): use the npm-ci-retry action on every install step

Fast Quality Gates failed on the recurring onnxruntime-node postinstall
ETIMEDOUT (Microsoft CDN 150.171.x.x) - the same transient flake that has
hit Vitest and dast-smoke today. Only the Build job used the retry action;
the other five jobs (Docs, Fast Quality Gates, Vitest, Unit Tests,
changelog) still ran a bare install and die on any CDN hiccup. Use the
existing retry action (3 attempts, exponential backoff) on every install
step for consistency.

* Merge branch 'release/v3.8.50' into fix/antigravity-project-autocreate

* test(fix): refresh expired alibaba quota sample validity and onnxruntime pin for v3.8.50 base

- alibaba-free-tier-quota-fetcher.test.ts: sample quotaValidityPeriod
  (2026-08-16 16:00 UTC) is in the past, making every quota entry classify
  as expired/not_capable; bump to 2028-01-01 UTC so the text/merge
  classification tests exercise the intended path again.
- optional-transformers-dependency.test.ts: onnxruntime-node pin assertion
  updated from ~1.24.3 to ~1.27.0 to match package.json (bumped by #10403);
  the regular-not-optional intent is unchanged.

* test(fix): widen modelsDevSync lastSync wait from 200ms default to 2000ms

The truthy-spellings loop asserted each enabled case completes its first
fetch within waitFor's 200ms default timeout, which trips under CI runner
load (observed on PR 10424 shard 2/4). Match the file's other lastSync
waits (2000ms) so the sync-completion assertion is load-tolerant.

---------

Co-authored-by: Rouzbeh <rqzbeh@users.noreply.github.com>
This commit is contained in:
Rouzbeh†
2026-08-17 14:53:41 +03:30
committed by GitHub
parent 24ef1dc3d4
commit 6ff2e7b2c2
12 changed files with 489 additions and 48 deletions

View File

@@ -28,7 +28,10 @@ import {
resolveAntigravityOutputCap,
} from "./antigravityOutputCap.ts";
export { MAX_ANTIGRAVITY_OUTPUT_TOKENS } from "./antigravityOutputCap.ts";
import { ensureAntigravityProjectAssigned } from "../services/antigravityProjectBootstrap.ts";
import {
ensureAntigravityProjectAssigned,
ANTIGRAVITY_REQUIRES_MANUAL_PROJECT,
} from "../services/antigravityProjectBootstrap.ts";
import { persistDiscoveredAntigravityProjectId } from "../services/antigravityProjectPersist.ts";
import { markAntigravityMissingCloudCodeProject } from "../services/antigravityProjectPersistence.ts";
import {
@@ -577,6 +580,7 @@ export class AntigravityExecutor extends BaseExecutor {
// its Google account already owns a Cloud Code project (the OAuth-time loadCodeAssist
// returned empty/transiently failed). Mirror the Cloud Code bootstrap to recover it
// here — the helper memoizes per access-token, so this is a one-time round-trip.
let requiresManualProject = false;
if (!projectId && credentials?.accessToken) {
const discovered = await ensureAntigravityProjectAssigned(
credentials.accessToken,
@@ -584,7 +588,7 @@ export class AntigravityExecutor extends BaseExecutor {
getAntigravityClientProfile(credentials),
signal
);
if (discovered) {
if (discovered && discovered !== ANTIGRAVITY_REQUIRES_MANUAL_PROJECT) {
projectId = discovered;
// #8491: persist the recovered id so it survives the next token refresh
// or process restart instead of being silently rediscovered every time.
@@ -594,10 +598,40 @@ export class AntigravityExecutor extends BaseExecutor {
credentials.providerSpecificData
);
}
requiresManualProject = discovered === ANTIGRAVITY_REQUIRES_MANUAL_PROJECT;
}
if (!projectId) {
markAntigravityMissingCloudCodeProject(credentials?.connectionId);
if (requiresManualProject) {
// Google no longer auto-creates GCP projects for standard-tier
// accounts (tracked in #8491): fail fast with a clear instruction
// instead of the generic 422 — a fabricated/omitted id only earns a
// delayed 429 RESOURCE_EXHAUSTED from Google's quota check.
const errorBody = {
error: {
message:
"GCP_PROJECT_REQUIRED: Google Antigravity now requires a free GCP Project ID. " +
"Create one at console.cloud.google.com and enter it in Providers → Antigravity " +
"(connection settings → Project ID). Automatic project creation is no longer " +
"available for personal accounts.",
type: "gcp_project_required",
code: "gcp_project_required",
},
};
// 422, not 403: chatCore's generic "401/403 → refresh credentials and
// retry" path would otherwise hit Google's OAuth token endpoint on
// every request from an affected account — pointless, since refreshing
// the token cannot create a GCP project. 422 also matches the sibling
// missing_project_id error, which the client already maps to a clear
// "action needed" prompt.
const resp = new Response(JSON.stringify(errorBody), {
status: 422,
headers: { "Content-Type": "application/json" },
});
// Returning a Response object signals the executor to stop and forward it
return resp as unknown as never;
}
// (#489) Return a structured error instead of throwing — gives the client a clear signal
// to show a "Reconnect OAuth" prompt rather than an opaque "Internal Server Error".
const errorMsg =

View File

@@ -20,7 +20,10 @@ import {
} from "./antigravityHeaders.ts";
import { extractCodeAssistOnboardTierId } from "./codeAssistSubscription.ts";
import type { AntigravityClientProfile } from "./antigravityClientProfile.ts";
import { ANTIGRAVITY_BOOTSTRAP_BASE_URLS, getAntigravityOnboardUrls } from "../config/antigravityUpstream.ts";
import {
ANTIGRAVITY_BOOTSTRAP_BASE_URLS,
getAntigravityOnboardUrls,
} from "../config/antigravityUpstream.ts";
const LOAD_CODE_ASSIST_PATH = "/v1internal:loadCodeAssist";
const BOOTSTRAP_TIMEOUT_MS = 8_000;
@@ -47,7 +50,39 @@ function evictOldest(cache: Map<string, unknown>): void {
const projectCache = new Map<string, string>();
/** Per-key lock to prevent concurrent onboard attempts for the same token. */
const onboardLocks = new Map<string, Promise<boolean>>();
const onboardLocks = new Map<string, Promise<void>>();
/**
* Sentinel returned by ensureAntigravityProjectAssigned when Google's
* onboardUser completed but did NOT return a project id — no automatic
* project creation for standard-tier (personal) accounts (tracked in #8491),
* so Google requires a user-defined GCP project (BYOP). The
* caller must fail fast with a clear "enter your GCP project id" error
* instead of retrying (a fabricated id gets a delayed 429 RESOURCE_EXHAUSTED).
*/
export const ANTIGRAVITY_REQUIRES_MANUAL_PROJECT = "__REQUIRES_GCP_PROJECT__";
/**
* Per-token cache of accounts Google told us to Bring Your Own Project.
* Permanent for the process lifetime (LRU-capped): re-running onboardUser
* for such an account is a pointless ~18s quota-check round-trip that
* always comes back empty. Cleared by clearAntigravityProjectCache(); a
* manually-entered project id (stored on the connection) short-circuits
* before this is consulted.
*/
const requiresManualProjectCache = new Set<string>();
function markRequiresManualProject(key: string): void {
if (requiresManualProjectCache.size >= MAX_CACHE_SIZE) {
const oldest = requiresManualProjectCache.values().next().value;
if (oldest !== undefined) requiresManualProjectCache.delete(oldest);
}
requiresManualProjectCache.add(key);
}
/** Outcome of an onboardUser attempt — three-way so the caller can distinguish
* "transient failure (retry later)" from "Google says bring your own project". */
type AntigravityOnboardStatus = "onboarded" | "requires_manual_project" | "failed";
type FetchLike = (url: string, init?: RequestInit) => Promise<Response>;
@@ -138,7 +173,7 @@ async function tryOnboardUser(
clientProfile: AntigravityClientProfile,
tierId: string,
signal?: AbortSignal
): Promise<boolean> {
): Promise<AntigravityOnboardStatus> {
const urls = getAntigravityOnboardUrls();
const headers = getAntigravityContentHeaders(clientProfile, accessToken);
@@ -157,7 +192,20 @@ async function tryOnboardUser(
});
if (response.ok) {
return true;
// Accounts Google expects to Bring Their Own Project: onboardUser
// returns 200 without a `cloudaicompanionProject` in the body — no
// automatic project creation for standard-tier/personal accounts
// (tracked in #8491). Detect that so we can fail fast with a clear
// instruction instead of retrying forever or fabricating an id that
// Google later rejects with a delayed 429 RESOURCE_EXHAUSTED.
const body = await response.text().catch(() => "");
if (body && !/cloudaicompanionProject/.test(body)) {
console.warn(
`[models] antigravity onboardUser done but no project in response at ${url} — Google BYOP (user-defined GCP project) required`
);
return "requires_manual_project";
}
return "onboarded";
}
console.warn(
@@ -171,18 +219,40 @@ async function tryOnboardUser(
console.warn(`[models] antigravity onboardUser threw for ${url}: ${msg} — trying next`);
}
}
return false;
return "failed";
}
/** Per-token memoization for accounts we already tried onboarding (avoid repeated calls). */
const onboardAttemptedCache = new Set<string>();
/**
* Per-token failure backoff for the onboardUser creation path.
*
* A FAILED onboard attempt must never be memoized as "done": a transient
* upstream/network error would otherwise poison the account for the whole
* process lifetime, so every later request 422s with "Missing Google
* projectId" even though onboarding would succeed on retry. Instead we record
* WHEN a failure happened and only skip re-attempts while the short backoff
* window is open — the account heals itself on the next request after it
* expires. Successful discoveries are memoized in `projectCache` (with LRU
* eviction) and clear any pending failure marker.
*/
const onboardFailureAt = new Map<string, number>();
const ONBOARD_RETRY_BACKOFF_MS = 5 * 60 * 1000;
function addToOnboardAttemptedCache(key: string): void {
if (onboardAttemptedCache.size >= MAX_CACHE_SIZE) {
const oldest = onboardAttemptedCache.values().next().value;
if (oldest !== undefined) onboardAttemptedCache.delete(oldest);
function markOnboardFailure(key: string): void {
if (onboardFailureAt.size >= MAX_CACHE_SIZE) {
const oldest = onboardFailureAt.keys().next().value;
if (oldest !== undefined) onboardFailureAt.delete(oldest);
}
onboardAttemptedCache.add(key);
onboardFailureAt.set(key, Date.now());
}
function isOnboardOnBackoff(key: string): boolean {
const failedAt = onboardFailureAt.get(key);
if (failedAt === undefined) return false;
if (Date.now() - failedAt >= ONBOARD_RETRY_BACKOFF_MS) {
onboardFailureAt.delete(key);
return false;
}
return true;
}
/**
@@ -212,49 +282,71 @@ export async function ensureAntigravityProjectAssigned(
}
const { projectId: initialProjectId, tierId } = await tryLoadCodeAssist(
accessToken, fetchImpl, clientProfile, signal
accessToken,
fetchImpl,
clientProfile,
signal
);
let projectId = initialProjectId;
// Google told us this account must Bring Its Own Project — fail fast with
// the sentinel instead of repeating the pointless ~18s onboard round-trip.
if (!projectId && requiresManualProjectCache.has(cacheKey)) {
return ANTIGRAVITY_REQUIRES_MANUAL_PROJECT;
}
// loadCodeAssist is read-only — if the account was never onboarded, it returns
// empty. Call onboardUser to create the project, then retry discovery.
if (!projectId && !onboardAttemptedCache.has(cacheKey)) {
// Re-attempts are bounded by a short failure backoff (not a permanent memo),
// so a transient onboard failure heals on the next request. Accounts Google
// marks BYOP are cached permanently and short-circuit above.
if (!projectId && !isOnboardOnBackoff(cacheKey)) {
// Per-key lock: concurrent calls for the same token share one onboard attempt.
let lock = onboardLocks.get(cacheKey);
if (!lock) {
lock = (async () => {
let aborted = false;
let succeeded = false;
let requiresManual = false;
try {
const onboarded = await tryOnboardUser(
accessToken, fetchImpl, clientProfile, tierId, signal
const status = await tryOnboardUser(
accessToken,
fetchImpl,
clientProfile,
tierId,
signal
);
if (onboarded) {
const retry = await tryLoadCodeAssist(
accessToken, fetchImpl, clientProfile, signal
);
if (status === "requires_manual_project") {
markRequiresManualProject(cacheKey);
requiresManual = true;
return;
}
if (status === "onboarded") {
const retry = await tryLoadCodeAssist(accessToken, fetchImpl, clientProfile, signal);
if (retry.projectId) {
evictOldest(projectCache);
projectCache.set(cacheKey, retry.projectId);
return true;
succeeded = true;
return;
}
}
return false;
} catch (e) {
aborted = signal?.aborted === true;
return false;
return;
} finally {
onboardLocks.delete(cacheKey);
if (!aborted) addToOnboardAttemptedCache(cacheKey);
if (!aborted && !requiresManual) {
if (succeeded) onboardFailureAt.delete(cacheKey);
else markOnboardFailure(cacheKey);
}
}
})();
onboardLocks.set(cacheKey, lock);
}
const success = await lock;
if (success) {
const cached = projectCache.get(cacheKey);
if (cached) return cached;
}
await lock;
if (projectCache.has(cacheKey)) return projectCache.get(cacheKey);
if (requiresManualProjectCache.has(cacheKey)) return ANTIGRAVITY_REQUIRES_MANUAL_PROJECT;
}
if (projectId) {
@@ -268,10 +360,17 @@ export async function ensureAntigravityProjectAssigned(
/** Exported for tests. */
export function clearAntigravityProjectCache(): void {
projectCache.clear();
onboardAttemptedCache.clear();
onboardFailureAt.clear();
requiresManualProjectCache.clear();
onboardLocks.clear();
}
/** Test-only: clear the onboard failure backoff (simulates backoff expiry). */
export function clearAntigravityOnboardBackoff(key?: string): void {
if (key) onboardFailureAt.delete(key);
else onboardFailureAt.clear();
}
/** Exported for tests — inspect cache state. */
export function getAntigravityProjectFromCache(
accessToken: string,

View File

@@ -336,6 +336,7 @@ async function _getAccessTokenInternal(provider, credentials, log, proxyConfig:
if (
result?.accessToken &&
(provider === "antigravity" || provider === "agy") &&
!credentials.providerSpecificData?.isProjectIdManual &&
!(credentials.projectId || credentials.providerSpecificData?.projectId)
) {
try {