@oh-my-pi/pi-ai 17.2.14 → 17.3.0

package/CHANGELOG.md

@@ -2,6 +2,41 @@
 
 ## [Unreleased]
 
+## [17.3.0] - 2026-08-13
+
+### Breaking Changes
+
+- Renamed `withGeminiThinkingLoopGuard` to `withThinkingLoopGuard`; the guard applies to Gemini, DeepSeek, and Grok model-id families.
+
+### Changed
+
+- Updated OpenCode Go integration to use the official usage endpoint, removing hardcoded caps, enabling real-time credential validation, and routing multi-key pools based on rolling and weekly headroom.
+- Optimized Anthropic prompt caching with rolling 5-minute breakpoints and idle refreshes to keep the prompt prefix warm.
+
+### Fixed
+
+- Fixed Ollama chat adapter to correctly forward sampling parameters like temperature and topP to the provider.
+- Fixed OpenAI agent turns ending prematurely after a web search with no visible answer, ensuring the agent continues processing the search results.
+- Fixed a resource leak where completed model streams retained provider concurrency permits longer than necessary.
+- Fixed image input support for qwen3.8-max and newer models when using DashScope compatible-mode.
+- Fixed xAI usage reporting falling back to a stale cache when a new weekly cycle starts with 0% consumed credits.
+- Fixed Together AI login validation failures by querying the authenticated models list instead of a hardcoded model.
+- Fixed credential-health probes and usage fetches failing when using reference-stored API keys (such as environment variables or commands) by ensuring secrets are correctly resolved.
+- Fixed Perplexity email-OTP login by preserving the session cookies required for verification.
+- Fixed thinking configuration for OpenAI and Daybreak models to correctly send reasoning.effort: "none" when thinking is disabled.
+- Fixed Grok runaway thinking streams bypassing the thinking-loop guard.
+
+### Removed
+
+- Removed legacy local request-cost estimation machinery and database schemas previously used for OpenCode Go estimates.
+
+## [17.2.15] - 2026-08-12
+
+### Fixed
+
+- Fixed an issue where AWS_BEDROCK_SKIP_AUTH failed to expose Amazon Bedrock models when AWS credential files were unavailable.
+- Fixed an issue where forceReasoningOff was ignored by Anthropic and Google transports, which allowed native thinking alongside a caller-supplied external scratchpad.
+
 ## [17.2.14] - 2026-08-11
 
 ### Added

package/dist/types/auth/sqlite-credential-store.d.ts

@@ -7,7 +7,7 @@
 import { Database } from "bun:sqlite";
 import type { AuthCredential, AuthCredentialStore, CredentialRefreshLeaseFence, DisabledCredentialSummary, StoredAuthCredential, StoredCredentialBlock } from "../auth-storage.js";
 import type { OAuthCredentials } from "../registry/oauth/types.js";
-import type { ClientUsageReport, ClientUsageSummary, UsageCostHistoryEntry, UsageCostHistoryQuery, UsageHistoryEntry, UsageHistoryQuery } from "../usage.js";
+import type { ClientUsageReport, ClientUsageSummary, UsageHistoryEntry, UsageHistoryQuery } from "../usage.js";
 export declare const USAGE_REPORT_TTL_MS: number;
 type SerializedCredentialRecord = {
     credentialType: AuthCredential["type"];
@@ -77,8 +77,6 @@ export declare class SqliteAuthCredentialStore implements AuthCredentialStore {
     releaseCredentialRefreshLease(credentialId: number, owner: string): void;
     recordUsageSnapshots(entries: UsageHistoryEntry[]): void;
     listUsageHistory(query?: UsageHistoryQuery): UsageHistoryEntry[];
-    recordUsageCosts(entries: UsageCostHistoryEntry[]): void;
-    listUsageCosts(query?: UsageCostHistoryQuery): UsageCostHistoryEntry[];
     recordClientUsage(report: ClientUsageReport): void;
     getClientUsageSummary(sinceMs: number): ClientUsageSummary;
     /**

package/dist/types/auth-storage.d.ts

@@ -1,7 +1,7 @@
 import type { ApiKeyResolver } from "./auth-retry.js";
 import type { OAuthAuthInfo, OAuthController, OAuthCredentials, OAuthProviderId } from "./registry/oauth/types.js";
 import type { Provider } from "./types.js";
-import type { ClientUsageReport, ClientUsageSummary, CredentialRankingStrategy, ObservedUsageEntry, UsageCostHistoryEntry, UsageCostHistoryQuery, UsageHistoryEntry, UsageHistoryQuery, UsageLogger, UsageProvider, UsageReport } from "./usage.js";
+import type { ClientUsageReport, ClientUsageSummary, CredentialRankingStrategy, ObservedUsageEntry, UsageHistoryEntry, UsageHistoryQuery, UsageLogger, UsageProvider, UsageReport } from "./usage.js";
 import { type CodexResetConsumeCode, type CodexResetCredit } from "./usage/openai-codex-reset.js";
 export { isSqliteBusyError, isSqliteCorruptionError, SqliteAuthCredentialStore, } from "./auth/sqlite-credential-store.js";
 export type ApiKeyCredential = {
@@ -311,10 +311,6 @@ export interface AuthCredentialStore {
      * skipped — the broker host records into its own database instead.
      */
     recordUsageSnapshots?(entries: UsageHistoryEntry[]): void;
-    /** Append observed request costs for providers without upstream usage APIs. */
-    recordUsageCosts?(entries: UsageCostHistoryEntry[]): void;
-    /** Read observed request costs, oldest first. */
-    listUsageCosts?(query?: UsageCostHistoryQuery): UsageCostHistoryEntry[];
     /** Read recorded usage-limit snapshots, oldest first. */
     listUsageHistory?(query?: UsageHistoryQuery): UsageHistoryEntry[];
     /**
@@ -883,12 +879,6 @@ export declare class AuthStorage {
      * store has no durable history (e.g. a broker-backed remote store).
      */
     listUsageHistory(query?: UsageHistoryQuery): UsageHistoryEntry[];
-    /** Record one observed provider request cost for later local usage aggregation. */
-    recordUsageCost(provider: Provider, costUsd: number, options?: {
-        sessionId?: string;
-        recordedAt?: number;
-        baseUrl?: string;
-    }): boolean;
     /**
      * Forward one completed request's usage to the store's observer hook.
      * Broker-backed stores batch these into per-install reports so the broker

package/dist/types/providers/anthropic-client.d.ts

@@ -1,7 +1,7 @@
 import { AnthropicApiError, AnthropicConnectionError, AnthropicConnectionTimeoutError } from "../error/index.js";
 export { AnthropicApiError, AnthropicConnectionError, AnthropicConnectionTimeoutError };
 import type { FetchImpl } from "../types.js";
-import type { MessageCreateParamsStreaming } from "./anthropic-wire.js";
+import type { MessageCreateParams } from "./anthropic-wire.js";
 /** Per-request options accepted by {@link AnthropicMessages.create}. */
 export interface AnthropicRequestOptions {
     signal?: AbortSignal;
@@ -78,7 +78,7 @@ export declare class AnthropicApiRequest {
 export declare class AnthropicMessages {
     #private;
     constructor(client: AnthropicMessagesClient, path: string);
-    create(params: MessageCreateParamsStreaming, options?: AnthropicRequestOptions): AnthropicApiRequest;
+    create(params: MessageCreateParams, options?: AnthropicRequestOptions): AnthropicApiRequest;
 }
 /**
  * Structural interface satisfied by both {@link AnthropicMessagesClient} and
@@ -87,11 +87,11 @@ export declare class AnthropicMessages {
  */
 export interface AnthropicMessagesClientLike {
     messages: {
-        create(params: MessageCreateParamsStreaming, options?: AnthropicRequestOptions): unknown;
+        create(params: MessageCreateParams, options?: AnthropicRequestOptions): unknown;
     };
     beta?: {
         messages: {
-            create(params: MessageCreateParamsStreaming, options?: AnthropicRequestOptions): unknown;
+            create(params: MessageCreateParams, options?: AnthropicRequestOptions): unknown;
         };
     };
 }
@@ -102,5 +102,5 @@ export declare class AnthropicMessagesClient implements AnthropicMessagesClientL
         readonly messages: AnthropicMessages;
     };
     constructor(options: AnthropicClientOptions);
-    request(path: string, params: MessageCreateParamsStreaming, options?: AnthropicRequestOptions): AnthropicApiRequest;
+    request(path: string, params: MessageCreateParams, options?: AnthropicRequestOptions): AnthropicApiRequest;
 }

package/dist/types/providers/anthropic.d.ts

@@ -1,6 +1,6 @@
 import type { Api, FetchImpl, Message, Model, ProviderSessionState, ServiceTier, SimpleStreamOptions, StreamFunction, StreamOptions, Usage } from "../types.js";
 import { type AnthropicFetchOptions, type AnthropicMessagesClientLike } from "./anthropic-client.js";
-import { type FallbackParam, type MessageParam, type TextBlockParam } from "./anthropic-wire.js";
+import { type FallbackParam, type MessageParam } from "./anthropic-wire.js";
 export type AnthropicHeaderOptions = {
     apiKey: string;
     baseUrl?: string;
@@ -17,7 +17,6 @@ export type AnthropicHeaderOptions = {
 export declare function normalizeAnthropicBaseUrl(baseUrl?: string): string | undefined;
 export declare function buildBetaHeader(baseBetas: readonly string[], extraBetas: readonly string[]): string;
 export declare function buildAnthropicHeaders(options: AnthropicHeaderOptions): Record<string, string>;
-type AnthropicCacheControl = NonNullable<TextBlockParam["cache_control"]>;
 /**
  * Clears the in-session "server rejected fast mode" sticky flag. Call when the
  * caller is explicitly re-arming `serviceTier: "priority"` (e.g. user toggled
@@ -229,14 +228,12 @@ export declare const streamAnthropic: StreamFunction<"anthropic-messages">;
 export type AnthropicSystemBlock = {
     type: "text";
     text: string;
-    cache_control?: AnthropicCacheControl;
 };
 type SystemBlockOptions = {
     includeClaudeCodeInstruction?: boolean;
     extraInstructions?: string[];
     /** Text of the first user message — used as fingerprint seed for the billing header. */
     firstUserMessageText?: string;
-    cacheControl?: AnthropicCacheControl;
 };
 export declare function buildAnthropicSystemBlocks(systemPrompt: readonly string[] | undefined, options?: SystemBlockOptions): AnthropicSystemBlock[] | undefined;
 export declare function normalizeExtraBetas(betas?: string[] | string): string[];

package/dist/types/providers/vision-guard.d.ts

@@ -12,7 +12,11 @@ export declare function joinTextWithImagePlaceholder(text: string, omittedImages
  * multimodal content arrays for. The compatible-mode endpoint also serves
  * multimodal Qwen SKUs without `vl` in the id (e.g. `qwen3.7-plus`), so this
  * guard only covers families verified to be text-only for issue #1859:
- * `qwen*-max` and `qwen*-coder*`.
+ * `qwen*-coder*` and `qwen*-max` up to and including `qwen3.7-max`.
+ *
+ * Qwen-Max became multimodal at `qwen3.8-max` (image input, issue #8019), so
+ * `-max` SKUs at version 3.8 or newer are excluded — otherwise the override
+ * would strip images from a genuinely vision-capable flagship (issue #8305).
  *
  * Used as a defensive override in `convertMessages` so a misconfigured custom
  * provider (issue #1859) can't drive the request into an unrecoverable 400.

package/dist/types/registry/amazon-bedrock.d.ts

@@ -1,8 +1,7 @@
-import { resolveAwsRegistryApiKey } from "./aws.js";
 export declare const amazonBedrockProvider: {
     readonly id: "amazon-bedrock";
     readonly name: "Amazon Bedrock";
-    readonly envKeys: typeof resolveAwsRegistryApiKey;
+    readonly envKeys: () => string | undefined;
     readonly mapSimpleOptions: (options: import("../index.js").SimpleStreamOptions) => {
         region: string | undefined;
         profile: string | undefined;

package/dist/types/registry/aws.d.ts

@@ -8,6 +8,8 @@ export interface AwsBedrockProviderOptions extends Readonly<Record<string, unkno
 }
 export declare function hasAwsCredentialSource(): boolean;
 /** Registry key marker for AWS transports that resolve their own bearer/IAM credentials. */
-export declare function resolveAwsRegistryApiKey(): string | undefined;
+export declare function resolveAwsRegistryApiKey(options?: {
+    allowSkipAuth?: boolean;
+}): string | undefined;
 /** Resolve a real AWS bearer token while filtering the registry's auth marker. */
 export declare function resolveAwsBearerToken(apiKey?: string, bearerToken?: string): string | undefined;

package/dist/types/registry/registry.d.ts

@@ -25,7 +25,7 @@ declare const ALL: ({
 } | {
     readonly id: "amazon-bedrock";
     readonly name: "Amazon Bedrock";
-    readonly envKeys: typeof import("./aws.js").resolveAwsRegistryApiKey;
+    readonly envKeys: () => string | undefined;
     readonly mapSimpleOptions: (options: import("../index.js").SimpleStreamOptions) => {
         region: string | undefined;
         profile: string | undefined;

package/dist/types/stream.d.ts

@@ -10,6 +10,12 @@ export declare function isOfficialCodexApiUrl(baseUrl: string | undefined): bool
 export declare function configureProviderMaxInFlightRequests(limits: Record<string, number> | undefined): void;
 export declare const __providerInFlightForTesting: {
     setRoot(root: string | undefined): void;
+    setHeartbeatTimings(timings: {
+        heartbeatMs?: number;
+        heartbeatFlushTimeoutMs?: number;
+    } | undefined): void;
+    setHeartbeatWriter(writer: ((writeProviderInFlightInfo: () => Promise<void>) => Promise<void>) | undefined): void;
+    setLeaseRemover(remover: ((leasePath: string) => Promise<void>) | undefined): void;
     providerDir(provider: string): string;
     lockDir(provider: string): string;
     captureStaleLockRelease(provider: string): Promise<(() => Promise<void>) | null>;

package/dist/types/types.d.ts

@@ -228,6 +228,16 @@ export interface StreamOptions {
     signal?: AbortSignal;
     apiKey?: string;
     cacheRetention?: CacheRetention;
+    /**
+     * Keep Anthropic's 5-minute prompt cache warm across bounded idle gaps.
+     *
+     * This is an ownership flag, not a general provider default: exactly one
+     * primary agent loop sharing `providerSessionState` should enable it.
+     * Side-channel and advisor requests must leave it unset.
+     */
+    anthropicCacheRefresh?: boolean;
+    /** @internal Marks a replay-only Anthropic request that must use non-streaming `max_tokens: 0`. */
+    anthropicCacheRefreshRequest?: boolean;
     /**
      * Additional headers to include in provider requests.
      * These are merged on top of model-defined headers.
@@ -295,8 +305,9 @@ export interface StreamOptions {
      */
     statefulResponses?: boolean;
     /**
-     * Emit `reasoning: { effort: "none" }` for OpenAI Responses and Codex requests.
-     * Used when a caller supplies an external reasoning scratchpad; other transports ignore it.
+     * Disable native reasoning when the caller supplies an external scratchpad.
+     * OpenAI Responses emits `reasoning: { effort: "none" }`; Anthropic and
+     * Google transports use their native thinking-off controls.
      */
     forceReasoningOff?: boolean;
     /**
@@ -374,10 +385,10 @@ export interface StreamOptions {
      */
     providerRetryWait?: (delayMs: number, signal?: AbortSignal) => Promise<void>;
     /**
-     * Accept a Google `STOP` response with no visible text or tool call as a
-     * successful completion. Passive callers such as advisors use this because
-     * silence is a valid result; interactive agent turns retain empty-response
-     * retries by default. Ignored by non-Google providers.
+     * Accept a normal provider stop with no visible text or tool call as a
+     * successful completion. Passive callers and zero-output cache refreshes use
+     * this because silence is their expected result; interactive agent turns
+     * retain empty-response retries by default.
      */
     acceptEmptyResponse?: boolean;
     /**

package/dist/types/usage/opencode-go.d.ts

@@ -1,2 +1,13 @@
-import type { UsageProvider } from "../usage.js";
+import type { CredentialRankingStrategy, UsageProvider } from "../usage.js";
 export declare const opencodeGoUsageProvider: UsageProvider;
+/**
+ * Multi-key pools rank by real headroom on the rolling and weekly windows.
+ *
+ * The monthly window is deliberately display-only: an exhausted monthly can
+ * still serve requests when the account's console "Use balance" fallback is
+ * enabled, and the usage endpoint does not report that flag — blocking on it
+ * would bench a working key until the subscription anniversary. Hard monthly
+ * failures still rotate credentials via the `401 Insufficient balance`
+ * usage-limit classification ([#3169](https://github.com/can1357/oh-my-pi/issues/3169)).
+ */
+export declare const opencodeGoRankingStrategy: CredentialRankingStrategy;

package/dist/types/usage.d.ts

@@ -139,23 +139,6 @@ export interface UsageHistoryQuery {
     /** Inclusive lower bound on {@link UsageHistoryEntry.recordedAt} (epoch ms). */
     sinceMs?: number;
 }
-/** One observed provider request cost, attributed to the credential that made it. */
-export interface UsageCostHistoryEntry {
-    /** Epoch ms the request completed. */
-    recordedAt: number;
-    provider: Provider;
-    /** Stable credential identity key (account/email/project/secret derived). */
-    accountKey: string;
-    /** Estimated request cost in USD. */
-    costUsd: number;
-}
-/** Filter for reading observed request costs. */
-export interface UsageCostHistoryQuery {
-    provider?: string;
-    accountKey?: string;
-    /** Inclusive lower bound on {@link UsageCostHistoryEntry.recordedAt} (epoch ms). */
-    sinceMs?: number;
-}
 /**
  * Aggregated request usage a client observed for one (provider, model) pair.
  * Clients fold every completed request into per-pair buckets and flush them to
@@ -468,8 +451,6 @@ export interface UsageFetchContext {
     fetch: FetchImpl;
     logger?: UsageLogger;
     retryWait?: (delayMs: number, signal?: AbortSignal) => Promise<void>;
-    /** Observed request-cost history for providers without upstream usage APIs. */
-    listUsageCosts?: (query?: UsageCostHistoryQuery) => UsageCostHistoryEntry[];
 }
 /** Provider implementation for fetching usage information. */
 export interface UsageProvider {

package/dist/types/utils/empty-completion-retry.d.ts

@@ -11,6 +11,7 @@ export declare function hasVisibleAssistantContent(message: AssistantMessage): b
 interface EmptyCompletionRetryOptions {
     signal?: AbortSignal;
     providerRetryWait?: (delayMs: number, signal?: AbortSignal) => Promise<void>;
+    acceptEmptyResponse?: boolean;
 }
 /**
  * Wrap a single-attempt provider stream with bounded empty-completion retries.

package/dist/types/utils/thinking-loop.d.ts

@@ -5,24 +5,13 @@ import { AssistantMessageEventStream } from "./event-stream.js";
  *  classifiers treat it as a transient (retryable) stop without bespoke rules. */
 export declare const THINKING_LOOP_ERROR_MARKER = "Thinking loop detected";
 /**
- * True when `model` is a Gemini model whose native thinking stream surfaces the
- * "thought summary" titles this module's header guard counts.
+ * True when `model.id` belongs to a family guarded for thinking/response loops:
+ * Gemini, DeepSeek, or Grok.
  *
- * OpenAI-compat transports can serve Gemini under an arbitrary provider/id, so they
- * carry the explicit `compat.enableGeminiThinkingLoopGuard` flag; direct Gemini
- * transports carry a clearly shaped id/provider, so a string match is sufficient.
- */
-export declare function isGeminiThinkingModel(model: Model<Api>): boolean;
-/**
- * True when `model` should be guarded for thinking/response loops (Gemini & DeepSeek).
- *
- * OpenAI-compat transports can serve Gemini or DeepSeek under an arbitrary provider/id.
- * Direct Gemini/DeepSeek transports carry a clearly shaped id/provider, so a string match
- * is sufficient.
+ * Model identity is derived only from its id; provider and compatibility metadata
+ * do not opt opaque aliases into the guard.
  */
 export declare function isLoopGuardedModel(model: Model<Api>, options?: StreamOptions): boolean;
-/** @deprecated Use isLoopGuardedModel instead. */
-export declare function isGeminiThinkingLoopModel(model: Model<Api>): boolean;
 /**
  * Stateful detector fed the streamed thinking deltas. `push` returns a
  * human-readable reason the first time a loop shape is recognized; the caller
@@ -81,7 +70,7 @@ export declare class GeminiHeaderRunDetector {
 /**
  * Wrap a provider stream with the loop guard. `controller` is the guard's own
  * abort handle: aborting it (after wiring it into the provider's signal via
- * {@link withGeminiThinkingLoopGuard}) tears down the upstream once a loop
+ * {@link withThinkingLoopGuard}) tears down the upstream once a loop
  * trips.
  */
 export declare function guardThinkingLoopStream(inner: AssistantMessageEventStream, model: Model<Api>, controller: AbortController, options?: StreamOptions): AssistantMessageEventStream;
@@ -93,7 +82,7 @@ export declare function guardThinkingLoopStream(inner: AssistantMessageEventStre
  * stall; bounding the re-samples and the final cook pass lives in the
  * result-awaiting caller.
  */
-export declare function withGeminiThinkingLoopGuard<O extends {
+export declare function withThinkingLoopGuard<O extends {
     signal?: AbortSignal;
     loopGuard?: {
         enabled?: boolean;

package/package.json

@@ -1,7 +1,7 @@
 {
 	"type": "module",
 	"name": "@oh-my-pi/pi-ai",
-	"version": "17.2.14",
+	"version": "17.3.0",
 	"description": "Unified LLM API with automatic model discovery and provider configuration",
 	"homepage": "https://omp.sh",
 	"author": "Can Boluk",
@@ -38,10 +38,10 @@
 	},
 	"dependencies": {
 		"@bufbuild/protobuf": "^2.12.1",
-		"@oh-my-pi/omptype": "17.2.14",
-		"@oh-my-pi/pi-catalog": "17.2.14",
-		"@oh-my-pi/pi-utils": "17.2.14",
-		"@oh-my-pi/pi-wire": "17.2.14"
+		"@oh-my-pi/omptype": "17.3.0",
+		"@oh-my-pi/pi-catalog": "17.3.0",
+		"@oh-my-pi/pi-utils": "17.3.0",
+		"@oh-my-pi/pi-wire": "17.3.0"
 	},
 	"devDependencies": {
 		"@bufbuild/protoc-gen-es": "^2.12.1",

package/src/auth/sqlite-credential-store.ts

@@ -25,8 +25,6 @@ import type {
 	ClientProviderUsage,
 	ClientUsageReport,
 	ClientUsageSummary,
-	UsageCostHistoryEntry,
-	UsageCostHistoryQuery,
 	UsageHistoryEntry,
 	UsageHistoryQuery,
 } from "../usage";
@@ -387,8 +385,6 @@ export class SqliteAuthCredentialStore implements AuthCredentialStore {
 	#releaseCredentialRefreshLeaseStmt: Statement;
 	#credentialBlockReconcileAfter: Map<string, number> = new Map();
 	#insertUsageHistoryStmt: Statement;
-	#insertUsageCostStmt: Statement;
-	#listUsageCostsStmt: Statement;
 	#lastUsageHistoryStmt: Statement;
 	#listUsageHistoryStmt: Statement;
 	#updateUsageHistoryStmt: Statement;
@@ -516,12 +512,6 @@ export class SqliteAuthCredentialStore implements AuthCredentialStore {
 		this.#listUsageHistoryStmt = this.#db.prepare(
 			"SELECT recorded_at, provider, account_key, email, account_id, limit_id, label, window_label, used_fraction, status, resets_at FROM usage_history WHERE recorded_at >= ? AND (? IS NULL OR provider = ?) ORDER BY recorded_at ASC",
 		);
-		this.#insertUsageCostStmt = this.#db.prepare(
-			"INSERT INTO usage_cost_history (recorded_at, provider, account_key, cost_usd) VALUES (?, ?, ?, ?)",
-		);
-		this.#listUsageCostsStmt = this.#db.prepare(
-			"SELECT recorded_at, provider, account_key, cost_usd FROM usage_cost_history WHERE recorded_at >= ? AND (? IS NULL OR provider = ?) AND (? IS NULL OR account_key = ?) ORDER BY recorded_at ASC",
-		);
 	}
 
 	static async open(dbPath: string = getAgentDbPath()): Promise<SqliteAuthCredentialStore> {
@@ -634,14 +624,6 @@ export class SqliteAuthCredentialStore implements AuthCredentialStore {
 				resets_at INTEGER
 			);
 			CREATE INDEX IF NOT EXISTS idx_usage_history_series ON usage_history(provider, account_key, limit_id, recorded_at);
-			CREATE TABLE IF NOT EXISTS usage_cost_history (
-				id INTEGER PRIMARY KEY AUTOINCREMENT,
-				recorded_at INTEGER NOT NULL,
-				provider TEXT NOT NULL,
-				account_key TEXT NOT NULL,
-				cost_usd REAL NOT NULL
-			);
-			CREATE INDEX IF NOT EXISTS idx_usage_cost_history_lookup ON usage_cost_history(provider, account_key, recorded_at);
 			CREATE INDEX IF NOT EXISTS idx_usage_history_recorded ON usage_history(recorded_at);
 			CREATE TABLE IF NOT EXISTS clients (
 				install_id TEXT PRIMARY KEY,
@@ -1769,42 +1751,6 @@ export class SqliteAuthCredentialStore implements AuthCredentialStore {
 			return [];
 		}
 	}
-	recordUsageCosts(entries: UsageCostHistoryEntry[]): void {
-		try {
-			for (const entry of entries) {
-				this.#insertUsageCostStmt.run(entry.recordedAt, entry.provider, entry.accountKey, entry.costUsd);
-			}
-		} catch {
-			// Cost history is best-effort; never break request persistence.
-		}
-	}
-
-	listUsageCosts(query?: UsageCostHistoryQuery): UsageCostHistoryEntry[] {
-		try {
-			const provider = query?.provider ?? null;
-			const accountKey = query?.accountKey ?? null;
-			const rows = this.#listUsageCostsStmt.all(
-				query?.sinceMs ?? 0,
-				provider,
-				provider,
-				accountKey,
-				accountKey,
-			) as Array<{
-				recorded_at: number;
-				provider: string;
-				account_key: string;
-				cost_usd: number;
-			}>;
-			return rows.map(row => ({
-				recordedAt: row.recorded_at,
-				provider: row.provider as Provider,
-				accountKey: row.account_key,
-				costUsd: row.cost_usd,
-			}));
-		} catch {
-			return [];
-		}
-	}
 
 	recordClientUsage(report: ClientUsageReport): void {
 		const now = Date.now();
@@ -2051,8 +1997,6 @@ export class SqliteAuthCredentialStore implements AuthCredentialStore {
 		this.#lastUsageHistoryStmt.finalize();
 		this.#listUsageHistoryStmt.finalize();
 		this.#updateUsageHistoryStmt.finalize();
-		this.#insertUsageCostStmt.finalize();
-		this.#listUsageCostsStmt.finalize();
 		this.#updateIfMatchesStmt.finalize();
 		this.#updateIfMatchesWithLeaseStmt.finalize();
 		this.#deleteIfMatchesWithLeaseStmt.finalize();