@cortexkit/pi-anthropic-auth 2.4.10 → 2.4.12

package/dist/native.d.ts

@@ -27,6 +27,9 @@ export interface PiVaultPollOptions {
     clearInterval?: typeof globalThis.clearInterval;
 }
 export declare class PiNativeRuntime {
+    private readonly pendingReports;
+    /** Join started failure reports, never model sends waiting for headers. */
+    drainReports(): Promise<void>;
     private readonly paths;
     private readonly holderRegistration;
     private readonly initialization;

package/dist/native.js

@@ -71,6 +71,12 @@ export function isNativeMigrationRequired(error) {
 /** How often Pi asks the vault for account changes, matching OpenCode's default. Model requests no longer rediscover a healthy roster, so this poll is how Pi notices a re-login or a newly added account. */
 export const PI_VAULT_POLL_INTERVAL_MS = 2000;
 export class PiNativeRuntime {
+    pendingReports = new Set();
+    /** Join started failure reports, never model sends waiting for headers. */
+    async drainReports() {
+        while (this.pendingReports.size)
+            await Promise.allSettled([...this.pendingReports]);
+    }
     paths;
     holderRegistration;
     initialization;
@@ -328,8 +334,18 @@ export class PiNativeRuntime {
         };
     }
     async report(served, status, source) {
-        if (served.source === 'vault' && status === 401)
-            await (await this.service()).vault.reportFailure(served.receipt, status, source);
+        if (served.source === 'vault' && status === 401) {
+            // An already-served receipt can be reported while shutdown drains it.
+            // Do not re-enter service admission after close has blocked new sends.
+            const pending = this.initialization.then((runtime) => runtime.vault.reportFailure(served.receipt, status, source));
+            this.pendingReports.add(pending);
+            try {
+                await pending;
+            }
+            finally {
+                this.pendingReports.delete(pending);
+            }
+        }
     }
     async markUsed(served, servingRuntime) {
         const runtime = servingRuntime ?? (await this.service());
@@ -350,6 +366,7 @@ export class PiNativeRuntime {
             await this.initialization.then(async (runtime) => {
                 await runtime.drainQuotaPublications();
                 await runtime.drainRoutingPublications();
+                await this.drainReports();
                 await runtime.close();
             }, () => { });
         })();

package/dist/stream.d.ts

@@ -1,5 +1,5 @@
-import { type ApiKeyAccount, CacheKeepManager, type MidConversationEffortTransition, type QuotaManagerOptions } from '@cortexkit/anthropic-auth-core';
-import { type Api, type AssistantMessageEventStream, type Context, type Model, type SimpleStreamOptions } from '@earendil-works/pi-ai';
+import { type ApiKeyAccount, CacheKeepManager, type MidConversationEffortTransition, type NativeAccountView, type QuotaManagerOptions } from '@cortexkit/anthropic-auth-core';
+import { type Api, type AssistantMessageEventStream, type Context, type Model, type SimpleStreamOptions, type Tool } from '@earendil-works/pi-ai';
 /** Set or clear the test replacement for the wait between quota usage polls. */
 export declare function __setPiQuotaCallPauseForTest(pause: QuotaManagerOptions['pauseBetweenQuotaCalls']): void;
 export declare function clearPiStickyRoutingSession(storagePath: string, sessionId: string): Promise<void>;
@@ -20,6 +20,22 @@ type AnthropicEvent = {
 export declare function buildExplicitBaseMessagesUrl(baseURL: string): URL;
 export declare function configureApiRouteHeaders(account: ApiKeyAccount, fastMode: boolean): Headers;
 export declare function parseSse(response: Response): AsyncGenerator<AnthropicEvent>;
+declare function sendAnthropicRequest(options: {
+    model: Model<Api>;
+    context: Context;
+    streamOptions?: SimpleStreamOptions;
+    apiAccount?: NativeAccountView;
+    storagePath: string;
+    oauthAccountId?: string;
+    route?: string;
+    effortTransitions?: readonly MidConversationEffortTransition[];
+    onResolvedTools?: (tools: Tool[]) => void;
+}, deps?: {
+    fetch?: typeof fetch;
+    cacheKeep?: CacheKeepManager;
+}): Promise<Response>;
+/** Exercise one native send with isolated transport and scheduling in tests. */
+export declare const __sendPiAnthropicRequestForTest: typeof sendAnthropicRequest;
 export declare function primaryResponseAllowsApiFallback(preflight: Response | string): boolean;
 /**
  * Classify a successful stream by its first event without a tee: the bytes

package/dist/stream.js

@@ -1,4 +1,4 @@
-import { applyClaudeCodeHeaders, applyCustomHeaders, CACHE_KEEP_EXTENDED_TTL_BETA, CacheKeepManager, CacheKeepSessionRegistry, createNativeVaultNoRouteResponse, createProviderErrorResponseScope, createSseFramer, createStickyNoRouteResponse, decideStickyQuotaFailure, decodeSseFrame, dumpDirectRequest, FAST_MODE_BETA, getCache1hPersistentMode, getClaudeCodeIdentityForVerifiedAccount, getDefaultCacheKeepRegistryDirectory, getFallbackReauthLabels, getQuotaCheckIntervalMs, getRequestGzipLatch, getRequestGzipLatchPath, getRoutingMode, getScopedQuotaWindowForModel, getThinkingPrefixMismatchBehavior, isCache1hPersistentlyEnabled, isCacheKeepHybridActive, isDumpPersistentlyEnabled, isFastModePersistentlyEnabled, isKillswitchEnabled, isNativeTransientAccountError, isOAuthAccount, isPermanentRefreshError, isValidApiBaseURL, killswitchPassesPolicy, logger, mergeAnthropicBetas, NativeAccountRefusal, nativeAccountCanServe, parseSseFrame, QuotaManager, quotaBackoffActive, quotaExhaustionFromHeaders, quotaSnapshotHasStandardWindows, quotaSnapshotModelScopeIsExhausted, quotaSnapshotPassesModelScope, quotaSnapshotPassesPolicy, STICKY_ROUTING_MAIN_ACCOUNT_ID, StickySessionRouter, sendViaRelay, sendWithRequestGzip, setDumpEnabled, shouldFallbackStatus, stickyQuotaSnapshotIsFresh, stickyRetryAfterWithJitter, stickyRouteFamilyForModel, } from '@cortexkit/anthropic-auth-core';
+import { applyClaudeCodeHeaders, applyCustomHeaders, CACHE_KEEP_EXTENDED_TTL_BETA, CacheKeepManager, CacheKeepSessionRegistry, createNativeVaultNoRouteResponse, createProviderErrorResponseScope, createSseFramer, createStickyNoRouteResponse, decideStickyQuotaFailure, decodeSseFrame, dumpDirectRequest, FAST_MODE_BETA, getCache1hPersistentMode, getClaudeCodeIdentityForVerifiedAccount, getDefaultCacheKeepRegistryDirectory, getFallbackReauthLabels, getQuotaCheckIntervalMs, getRequestGzipLatch, getRequestGzipLatchPath, getRoutingMode, getScopedQuotaWindowForModel, getThinkingPrefixMismatchBehavior, isCache1hPersistentlyEnabled, isCacheKeepHybridActive, isDumpPersistentlyEnabled, isFastModePersistentlyEnabled, isKillswitchEnabled, isNativeTransientAccountError, isOAuthAccount, isPermanentRefreshError, isValidApiBaseURL, killswitchPassesPolicy, logger, mergeAnthropicBetas, NativeAccountRefusal, nativeAccountCanServe, parseSseFrame, QuotaManager, quotaBackoffActive, quotaExhaustionFromHeaders, quotaSnapshotHasStandardWindows, quotaSnapshotModelScopeIsExhausted, quotaSnapshotPassesModelScope, quotaSnapshotPassesPolicy, runWithRequestSignal, STICKY_ROUTING_MAIN_ACCOUNT_ID, StickySessionRouter, sendViaRelay, sendWithRequestGzip, setDumpEnabled, shouldFallbackStatus, stickyQuotaSnapshotIsFresh, stickyRetryAfterWithJitter, stickyRouteFamilyForModel, } from '@cortexkit/anthropic-auth-core';
 import { calculateCost, createAssistantMessageEventStream, } from '@earendil-works/pi-ai';
 import { buildAnthropicRequest, fromClaudeCodeToolName } from './convert.js';
 import { getPiNativeRuntime, piAttemptAccessToken, } from './native.js';
@@ -335,7 +335,9 @@ export async function* parseSse(response) {
         reader.releaseLock();
     }
 }
-async function sendAnthropicRequest(options) {
+async function sendAnthropicRequest(options, deps = {}) {
+    const fetchImpl = deps.fetch ?? globalThis.fetch;
+    const keep = deps.cacheKeep ?? cacheKeepManager;
     const native = getPiNativeRuntime(options.storagePath);
     // Capture the serving service while authorization is open. Shutdown rejects
     // new serving work immediately but drains writes for already-served attempts.
@@ -385,8 +387,10 @@ async function sendAnthropicRequest(options) {
         signal: options.streamOptions?.signal,
         ...(!options.apiAccount && { redirect: 'error' }),
     };
-    if (!options.apiAccount)
-        await cacheKeepManager.track({
+    const trackAcceptedRequest = () => {
+        if (options.apiAccount || init.signal?.aborted)
+            return;
+        keep.track({
             sessionId: relayAffinity,
             url: input.toString(),
             headers,
@@ -397,6 +401,7 @@ async function sendAnthropicRequest(options) {
             oauthAccountIdentity: verifiedUuid,
             accountStoragePath: options.storagePath,
         });
+    };
     const reported = new Set();
     let retried401 = false;
     let lastTransport;
@@ -455,13 +460,13 @@ async function sendAnthropicRequest(options) {
             // OAuth bodies go gzipped as Claude Code sends them; `bodyText` stays
             // the signed identity text for dumps and CacheKeep.
             if (options.apiAccount || typeof init.body !== 'string')
-                return scope.redactResponse(await fetch(input, init));
+                return scope.redactResponse(await fetchImpl(input, init));
             return scope.redactResponse(await sendWithRequestGzip({
                 url: input,
                 body: init.body,
                 headers,
                 latch: getRequestGzipLatch(getRequestGzipLatchPath(options.storagePath)),
-                send: (wireBody, wireHeaders) => fetch(input, { ...init, body: wireBody, headers: wireHeaders }),
+                send: (wireBody, wireHeaders) => fetchImpl(input, { ...init, body: wireBody, headers: wireHeaders }),
             }));
         };
         let response;
@@ -537,8 +542,10 @@ async function sendAnthropicRequest(options) {
             const subject = apiSubject;
             native.usagePublications.publish(() => servingRuntime.publishApi(subject, { lastUsed: Date.now() }));
         }
-        if (attempt && response.ok)
+        if (attempt && response.ok) {
+            trackAcceptedRequest();
             markServed(attempt);
+        }
         return response;
     };
     if (options.apiAccount)
@@ -601,10 +608,14 @@ async function sendAnthropicRequest(options) {
     // If the request was cancelled since, end it as cancelled here.
     if (relay401Attempt && response.status === 401)
         await endIfCancelled(response, init.signal);
-    if (response.ok && lastTransport === 'relay' && relayAttempt)
+    if (response.ok && lastTransport === 'relay' && relayAttempt) {
+        trackAcceptedRequest();
         markServed(relayAttempt);
+    }
     return response;
 }
+/** Exercise one native send with isolated transport and scheduling in tests. */
+export const __sendPiAnthropicRequestForTest = sendAnthropicRequest;
 /**
  * If the request's signal is aborted, cancels the response body and throws the
  * signal's abort error; otherwise does nothing. Call it before a rejected
@@ -1242,7 +1253,7 @@ export function streamCortexKitAnthropic(model, context, options, effortTransiti
             buildPiOAuthMessagesUrl(model.baseUrl);
             const storagePath = getPiAccountStoragePath();
             let hostTools = [];
-            const response = await executeWithFallback({
+            const response = await runWithRequestSignal(options?.signal, () => executeWithFallback({
                 model,
                 context,
                 streamOptions: options,
@@ -1251,7 +1262,7 @@ export function streamCortexKitAnthropic(model, context, options, effortTransiti
                 onResolvedTools: (tools) => {
                     hostTools = tools;
                 },
-            });
+            }));
             if (!response.ok) {
                 throw new Error(`Anthropic request failed: HTTP ${response.status} ${await response.text()}`);
             }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@cortexkit/pi-anthropic-auth",
-  "version": "2.4.10",
+  "version": "2.4.12",
   "type": "module",
   "repository": {
     "type": "git",
@@ -36,7 +36,7 @@
     "prepublishOnly": "bun run build"
   },
   "dependencies": {
-    "@cortexkit/anthropic-auth-core": "2.4.10"
+    "@cortexkit/anthropic-auth-core": "2.4.12"
   },
   "peerDependencies": {
     "@earendil-works/pi-ai": ">=0.86.1 <2",