@oh-my-pi/pi-ai 17.0.0 → 17.0.2

package/CHANGELOG.md

@@ -2,6 +2,34 @@
 
 ## [Unreleased]
 
+## [17.0.2] - 2026-07-17
+
+### Fixed
+
+- Automatically invalidate and rotate OAuth credentials when an "invalidated oauth token" error occurs.
+- Fixed auth-broker snapshot validation rejecting API keys stored via the `/login` flow, restoring support for gateway/broker setups serving login-sourced keys on custom hosts.
+- Fixed an issue where literal reasoning tags (e.g., `<think>`) inside Markdown code blocks or inline code were incorrectly treated as reasoning boundaries, which corrupted the rendered Markdown.
+- Classified HTTP 402 and "balance exhausted" quota responses as persistent usage limits, enabling automatic rotation of multi-account requests to a sibling credential.
+- Fixed `kimi-code` Anthropic-format requests ignoring custom provider base URLs.
+- Fixed an issue where GPT-5.6 Codex Responses-Lite requests failed with an HTTP 400 error due to invalid `tool_choice` parameters after tools were rewritten, by automatically downgrading forced hosted choices to `tool_choice: "auto"` while preserving explicit tool-use constraints.
+- Fixed Cursor streams prematurely reporting success before late CONNECT or gRPC terminal failures were observed, and resolved issues rejecting transport ends without a `turnEnded` signal.
+
+## [17.0.1] - 2026-07-16
+
+### Fixed
+
+- Fixed OpenRouter cost reporting to use the provider's authoritative account charge instead of catalog token-price estimates on both Responses and Chat Completions streams.
+- Fixed OpenAI Responses and Chat Completions requests forwarding unsupported sampling parameters such as `temperature` to o-series and GPT-5+ models, preventing 400 errors for mnemopi memory calls through GitHub Copilot GPT-5.6 Luna. ([#5606](https://github.com/can1357/oh-my-pi/issues/5606))
+- Fixed boolean JSON Schema subschemas (`true`/`false`) in MCP tool inputs triggering `400 INVALID_ARGUMENT` on the Google/Cloud Code Assist (Antigravity) transport by coercing them to their object equivalents (`true` → `{}`, `false` → `{ not: {} }`) before sending ([#5604](https://github.com/can1357/oh-my-pi/issues/5604)).
+- Fixed thinking-enabled Claude requests routed to `google-vertex` sending the `effort-2025-11-24` beta as an `anthropic-beta` HTTP header, which Vertex rawPredict rejects with a 400. The effort beta and the `output_config.effort` field are now gated off the Vertex path the same way `context-management-2025-06-27` already is ([#5614](https://github.com/can1357/oh-my-pi/issues/5614)).
+- Fixed custom and Foundry-routed Anthropic endpoints receiving first-party eager/legacy tool-streaming controls ([#5572](https://github.com/can1357/oh-my-pi/issues/5572)).
+- Parsed Ollama NDJSON response bytes directly instead of decoding and buffering every network chunk as text. ([#5542](https://github.com/can1357/oh-my-pi/issues/5542))
+- Fixed Amazon Bedrock stream error handling for non-`Error` values that `JSON.stringify` cannot serialize ([#5539](https://github.com/can1357/oh-my-pi/issues/5539)).
+- Fixed concurrent provider OAuth refreshes by serializing rotating-token updates across processes, fencing stale writes, and preventing background usage probes from disabling otherwise usable credentials ([#5396](https://github.com/can1357/oh-my-pi/issues/5396)).
+- Fixed OpenAI Codex WebSocket connections ignoring `PI_PROXY`, provider-specific proxy settings, and standard HTTPS/ALL proxy variables ([#5384](https://github.com/can1357/oh-my-pi/issues/5384)).
+- Fixed Anthropic account quota exhaustion (`This request would exceed your account's monthly spend limit`) hanging until the local deadline instead of surfacing the error: the `rate_limit_error` "spend limit" wording is now classified as a persistent usage limit, so it fails fast and rotates to a sibling credential rather than looping in the provider retry backoff. ([#4787](https://github.com/can1357/oh-my-pi/issues/4787))
+- Fixed OpenRouter daily free-model allowance errors (`free-models-per-day`) being treated as transient rate limits, so requests rotate from an exhausted API key to a healthy sibling credential. ([#4832](https://github.com/can1357/oh-my-pi/issues/4832))
+
 ## [17.0.0] - 2026-07-15
 
 ### Changed

package/dist/types/auth-broker/wire-schemas.d.ts

@@ -29,6 +29,7 @@ export declare const remoteOauthCredentialSchema: import("arktype/internal/varia
 export declare const apiKeyCredentialSchema: import("arktype/internal/variants/object.ts").ObjectType<{
     type: "api_key";
     key: string;
+    source?: "login" | undefined;
 }, {}>;
 /** Discriminated union accepted on POST /v1/credential (writes). */
 export declare const writableAuthCredentialSchema: import("arktype/internal/variants/object.ts").ObjectType<{
@@ -46,6 +47,7 @@ export declare const writableAuthCredentialSchema: import("arktype/internal/vari
 } | {
     type: "api_key";
     key: string;
+    source?: "login" | undefined;
 }, {}>;
 /** Discriminated union returned in snapshots (refresh is sentinel for OAuth). */
 export declare const snapshotCredentialSchema: import("arktype/internal/variants/object.ts").ObjectType<{
@@ -63,6 +65,7 @@ export declare const snapshotCredentialSchema: import("arktype/internal/variants
 } | {
     type: "api_key";
     key: string;
+    source?: "login" | undefined;
 }, {}>;
 export declare const credentialSnapshotEntrySchema: import("arktype/internal/variants/object.ts").ObjectType<{
     id: number;
@@ -82,6 +85,7 @@ export declare const credentialSnapshotEntrySchema: import("arktype/internal/var
     } | {
         type: "api_key";
         key: string;
+        source?: "login" | undefined;
     };
     identityKey: string | null;
 }, {}>;
@@ -109,6 +113,7 @@ export declare const snapshotEntrySchema: import("arktype/internal/variants/obje
     } | {
         type: "api_key";
         key: string;
+        source?: "login" | undefined;
     };
     identityKey: string | null;
     rotatesInMs: number | null;
@@ -153,6 +158,7 @@ export declare const snapshotResponseSchema: import("arktype/internal/variants/o
         } | {
             type: "api_key";
             key: string;
+            source?: "login" | undefined;
         };
         identityKey: string | null;
         rotatesInMs: number | null;
@@ -193,6 +199,7 @@ export declare const snapshotStreamSnapshotEventSchema: import("arktype/internal
         } | {
             type: "api_key";
             key: string;
+            source?: "login" | undefined;
         };
         identityKey: string | null;
         rotatesInMs: number | null;
@@ -234,6 +241,7 @@ export declare const snapshotStreamEntryEventSchema: import("arktype/internal/va
         } | {
             type: "api_key";
             key: string;
+            source?: "login" | undefined;
         };
         identityKey: string | null;
         rotatesInMs: number | null;
@@ -287,6 +295,7 @@ export declare const snapshotStreamEventSchema: import("arktype/internal/variant
         } | {
             type: "api_key";
             key: string;
+            source?: "login" | undefined;
         };
         identityKey: string | null;
         rotatesInMs: number | null;
@@ -326,6 +335,7 @@ export declare const snapshotStreamEventSchema: import("arktype/internal/variant
         } | {
             type: "api_key";
             key: string;
+            source?: "login" | undefined;
         };
         identityKey: string | null;
         rotatesInMs: number | null;
@@ -427,6 +437,7 @@ export declare const credentialRefreshResponseSchema: import("arktype/internal/v
         } | {
             type: "api_key";
             key: string;
+            source?: "login" | undefined;
         };
         identityKey: string | null;
     };
@@ -469,6 +480,7 @@ export declare const credentialUploadRequestSchema: import("arktype/internal/var
     } | {
         type: "api_key";
         key: string;
+        source?: "login" | undefined;
     };
 }, {}>;
 export declare const credentialUploadResponseSchema: import("arktype/internal/variants/object.ts").ObjectType<{
@@ -490,6 +502,7 @@ export declare const credentialUploadResponseSchema: import("arktype/internal/va
         } | {
             type: "api_key";
             key: string;
+            source?: "login" | undefined;
         };
         identityKey: string | null;
     }[];

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

@@ -559,6 +559,8 @@ export interface InvalidateCredentialMatchingOptions {
 }
 /** Options for refreshing one stored OAuth row through durable ownership. */
 export interface StoredOAuthRefreshOptions<T extends OAuthCredential = OAuthCredential> {
+    /** Stable row id when a provider has multiple OAuth credentials. */
+    credentialId?: number;
     observedCredential?: T;
     credentialFromRow: (credential: OAuthCredential) => T | undefined;
     forceRefresh?: boolean;

package/dist/types/error/auth-classify.d.ts

@@ -5,6 +5,8 @@
  * `@oh-my-pi/pi-ai` entrypoint name used by the coding agent and auth-broker.
  */
 export declare function isDefinitiveOAuthFailure(errorMsg: string): boolean;
+/** Whether an upstream response explicitly says the supplied OAuth bearer was invalidated. */
+export declare function isInvalidatedOAuthTokenError(error: unknown): boolean;
 /**
  * Whether an upstream failure should rotate to a sibling credential: a hard
  * `401`, a body-classified usage limit (Codex `usage_limit_reached`, Anthropic

package/dist/types/error/rate-limit.d.ts

@@ -20,6 +20,9 @@ export declare function calculateRateLimitBackoffMs(reason: RateLimitReason): nu
 /**
  * HTTP status codes that, absent richer body classification, represent an
  * account-local usage cap rather than a bad credential or a transient blip.
+ * HTTP 402 Payment Required is categorically an account-billing cap (xAI
+ * Grok Build "usage balance exhausted", DeepSeek "Insufficient Balance",
+ * OpenRouter credit exhaustion) — never a transient blip or bad credential.
  * Always combine with {@link isUsageLimitOutcome} when a message is available
  * — a 429 carrying transient rate-limit wording is NOT a usage cap.
  */
@@ -31,7 +34,7 @@ export declare function isUsageLimitStatus(status: number | undefined): boolean;
  *  1. Body matches {@link isUsageLimitError} (Codex `usage_limit_reached`,
  *     Anthropic account rate-limit, Google `resource_exhausted`, OpenAI
  *     `insufficient_quota`, …) → rotate.
- *  2. Status is not 429 → backoff (caller's domain).
+ *  2. Status is not a usage-limit status (429/402) → backoff (caller's domain).
  *  3. Body is absent or {@link isOpaqueStatusBody opaque} (just the status,
  *     empty JSON, HTTP framing only) → rotate conservatively: the server
  *     gave us nothing else to go on.
@@ -44,10 +47,11 @@ export declare function isUsageLimitStatus(status: number | undefined): boolean;
  */
 export declare function isUsageLimitOutcome(status: number | undefined, message: string | undefined): boolean;
 /**
- * A 429 body is opaque when it carries no signal beyond the status itself —
- * empty, whitespace-only, the status digits with HTTP/JSON framing, or
- * generic punctuation. Anything else (retry hints, capacity wording, error
- * descriptions) is informative enough to defer to the classifier.
+ * A usage-limit status body is opaque when it carries no signal beyond the
+ * status itself — empty, whitespace-only, the status digits with HTTP/JSON
+ * framing, or generic punctuation. Anything else (retry hints, capacity
+ * wording, error descriptions) is informative enough to defer to the
+ * classifier.
  */
 export declare function isOpaqueStatusBody(message: string): boolean;
 /**

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

@@ -29,6 +29,8 @@ export interface BlockState {
         [kStreamingBlockIndex]: number;
     }) | null;
     currentToolCall: ToolCallState | null;
+    /** MCP call IDs executed through Cursor's exec channel before their stream block arrives. */
+    resolvedMcpToolCallIds: Set<string>;
     firstTokenTime: number | undefined;
     setTextBlock: (b: (TextContent & {
         [kStreamingBlockIndex]: number;

package/dist/types/providers/openai-codex/request-transformer.d.ts

@@ -77,6 +77,7 @@ export declare function resolveCodexResponsesLite(model: Model<"openai-codex-res
 export interface CodexLiteShapedBody {
     instructions?: unknown;
     tools?: unknown;
+    tool_choice?: unknown;
     input?: unknown;
     parallel_tool_calls?: unknown;
 }
@@ -85,9 +86,14 @@ export interface CodexLiteShapedBody {
  * `build_responses_request` with `use_responses_lite`): strips pinned image
  * detail, forces parallel tool calling off, moves tools into a leading
  * `additional_tools` developer item and the base instructions into a
- * developer message, then omits top-level `instructions`/`tools`. Shared by
- * normal turns and both remote-compaction paths — codex-rs routes
- * `/responses/compact` through the same builder.
+ * developer message, then omits top-level `instructions`/`tools`. Because the
+ * rewrite removes top-level `tools`, a forced hosted-tool choice (e.g.
+ * `{ type: "web_search" }`) would leave the backend unable to validate the
+ * choice against a tools collection and it rejects the request with HTTP 400
+ * (#5771). Such choices must fall back to `"auto"`; explicit string constraints
+ * such as `"none"` and `"required"` remain valid. Shared by normal turns and
+ * both remote-compaction paths — codex-rs routes `/responses/compact` through
+ * the same builder.
  */
 export declare function applyCodexResponsesLiteShape(body: CodexLiteShapedBody): void;
 export declare function transformRequestBody(body: RequestBody, model: Model<"openai-codex-responses">, options?: CodexRequestOptions, prompt?: {

package/dist/types/providers/openai-shared.d.ts

@@ -68,6 +68,8 @@ export declare function applyOpenAIServiceTier(params: {
  * proxy can never skew those costs.
  */
 export declare function applyOpenAIResponsesServiceTierCost(model: Pick<Model, "provider">, usage: AssistantMessage["usage"], responseServiceTier: unknown, requestServiceTier: ServiceTier | null | undefined): void;
+/** Reconcile token-price estimates with OpenRouter's authoritative account charge. */
+export declare function applyOpenRouterReportedCost(model: Pick<Model, "provider">, usage: Usage, rawUsage: unknown): void;
 export interface OpenAIUsageAccountingInput {
     promptTokens: number;
     outputTokens: number;
@@ -295,12 +297,14 @@ export declare function applyChatCompletionsReasoningParams(params: OpenAIComple
 }) | undefined): void;
 export declare function disableChatCompletionsReasoningForDialect(params: OpenAICompletionsParams, compat: ResolvedOpenAICompat): void;
 /**
- * Output-token clamp for the Z.AI/GLM-5.2 reasoning dialect: these hosts accept
- * the full model window on reasoning turns, so clamp to the model cap. Returns
- * `undefined` for every other model, leaving {@link resolveOpenAIOutputTokenParam}
- * on its default `OPENAI_MAX_OUTPUT_TOKENS` clamp.
+ * Provider-specific Chat Completions output clamp.
+ *
+ * Most OpenAI-compatible endpoints retain the conservative 64k ceiling from
+ * {@link resolveOpenAIOutputTokenParam}. Z.AI/GLM-5.2 reasoning and native
+ * Moonshot K3 explicitly accept their full advertised model caps, so those
+ * routes clamp to `model.maxTokens` instead.
  */
-export declare function resolveZaiReasoningOutputClamp(model: Model<"openai-completions">, compat: ResolvedOpenAICompat): number | undefined;
+export declare function resolveOpenAICompletionsOutputClamp(model: Model<"openai-completions">, compat: ResolvedOpenAICompat): number | undefined;
 /**
  * Enable `tool_stream` for Z.AI/GLM-5.2 reasoning models when tools are present
  * (GLM-5.2 streams tool-call arguments incrementally and needs the flag to do so).
@@ -500,7 +504,9 @@ type CommonSamplingOptions = Pick<StreamOptions, "temperature" | "topP" | "topK"
  * can let the upstream apply its own default instead of 400-ing on `maxTokens` values that
  * reflect the model's context window rather than the upstream output limit.
  */
-export declare function applyCommonResponsesSamplingParams<P extends CommonResponsesParams>(params: P, options: CommonSamplingOptions | undefined, model: Pick<Model, "provider" | "api" | "id" | "omitMaxOutputTokens" | "maxTokens">): void;
+export declare function applyCommonResponsesSamplingParams<P extends CommonResponsesParams>(params: P, options: CommonSamplingOptions | undefined, model: Pick<Model, "provider" | "api" | "id" | "omitMaxOutputTokens" | "maxTokens"> & {
+    compat: Pick<ResolvedOpenAISharedCompat, "supportsSamplingParams">;
+}): void;
 type ReasoningOptions = {
     reasoning?: string;
     reasoningSummary?: "auto" | "detailed" | "concise" | null;

package/dist/types/providers/transform-messages.d.ts

@@ -1,11 +1,7 @@
 import type { Api, AssistantMessage, Message, Model } from "../types.js";
-/**
- * Normalize tool call ID for cross-provider compatibility.
- * OpenAI Responses API generates IDs that are 450+ chars with special characters like `|`.
- * Anthropic APIs require IDs matching ^[a-zA-Z0-9_-]+$ (max 64 chars).
- *
- * For aborted/errored turns, this function:
- * - Preserves tool call structure (unlike converting to text summaries)
- * - Injects synthetic "aborted" tool results
- */
+export declare function redactSensitiveCredentials(text: string): string;
+export declare function redactSensitiveInObject(val: unknown): {
+    result: unknown;
+    changed: boolean;
+};
 export declare function transformMessages<TApi extends Api>(messages: Message[], model: Model<TApi>, normalizeToolCallId?: (id: string, model: Model<TApi>, source: AssistantMessage) => string, maxNormalizedToolCallIdLength?: number, duplicateToolCallIdSuffixPrefix?: string, targetCompat?: Model<TApi>["compat"]): Message[];

package/dist/types/types.d.ts

@@ -525,7 +525,7 @@ export interface ContextSnapshot {
 }
 export interface AssistantMessage {
     role: "assistant";
-    content: (TextContent | ThinkingContent | RedactedThinkingContent | AnthropicFallbackContent | ToolCall)[];
+    content: (TextContent | ThinkingContent | RedactedThinkingContent | AnthropicFallbackContent | ImageContent | ToolCall)[];
     api: Api;
     provider: Provider;
     model: string;
@@ -716,6 +716,11 @@ export type AssistantMessageEvent = {
     contentIndex: number;
     content: string;
     partial: AssistantMessage;
+} | {
+    type: "image_end";
+    contentIndex: number;
+    content: ImageContent;
+    partial: AssistantMessage;
 } | {
     type: "toolcall_start";
     contentIndex: number;

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

@@ -3,9 +3,9 @@ import { AssistantMessageEventStream } from "./event-stream.js";
 export declare const MAX_EMPTY_COMPLETION_RETRIES = 2;
 export declare const EMPTY_COMPLETION_BASE_DELAY_MS = 500;
 /**
- * Whether a completed assistant message carries content worth delivering: a tool
- * call or any non-whitespace text. An empty/whitespace-only message — or one
- * that only ever produced thinking — is the "empty response" failure.
+ * Whether a completed assistant message carries content worth delivering: an
+ * image, tool call, or any non-whitespace text. An empty/whitespace-only message
+ * — or one that only ever produced thinking — is the "empty response" failure.
  */
 export declare function hasVisibleAssistantContent(message: AssistantMessage): boolean;
 interface EmptyCompletionRetryOptions {

package/dist/types/utils/schema/normalize.d.ts

@@ -1,7 +1,9 @@
 import { type DescriptionSpillFormat } from "./spill.js";
 import { type JsonObject } from "./types.js";
-export type ResidualSchemaIncompatibility = "type-array" | "type-null" | "nullable" | "combiners";
+export type ResidualSchemaIncompatibility = "type-array" | "type-null" | "nullable" | "combiners" | "not";
 export interface NormalizeSchemaOptions {
+    /** Coerce JSON Schema boolean subschemas for providers whose wire cannot encode them. */
+    coerceBooleanSubschemas?: boolean;
     unsupportedFields: (key: string) => boolean;
     normalizeFieldNames: boolean;
     collapseNullFields: boolean;

package/package.json

@@ -1,7 +1,7 @@
 {
 	"type": "module",
 	"name": "@oh-my-pi/pi-ai",
-	"version": "17.0.0",
+	"version": "17.0.2",
 	"description": "Unified LLM API with automatic model discovery and provider configuration",
 	"homepage": "https://omp.sh",
 	"author": "Can Boluk",
@@ -38,9 +38,9 @@
 	},
 	"dependencies": {
 		"@bufbuild/protobuf": "^2.12.1",
-		"@oh-my-pi/pi-catalog": "17.0.0",
-		"@oh-my-pi/pi-utils": "17.0.0",
-		"@oh-my-pi/pi-wire": "17.0.0",
+		"@oh-my-pi/pi-catalog": "17.0.2",
+		"@oh-my-pi/pi-utils": "17.0.2",
+		"@oh-my-pi/pi-wire": "17.0.2",
 		"arktype": "2.2.3",
 		"zod": "^4"
 	},

package/src/auth-broker/wire-schemas.ts

@@ -55,6 +55,7 @@ export const apiKeyCredentialSchema = type({
 	"+": "reject",
 	type: "'api_key'",
 	key: type("string").atLeastLength(1),
+	"source?": "'login'",
 });
 
 /** Discriminated union accepted on POST /v1/credential (writes). */

package/src/auth-retry.ts

@@ -1,6 +1,6 @@
 import type { OAuthAccess } from "./auth-storage";
 import * as AIError from "./error";
-import { isAuthRetryableError } from "./error/auth-classify";
+import { isAuthRetryableError, isInvalidatedOAuthTokenError } from "./error/auth-classify";
 import { isUsageLimit } from "./error/flags";
 import { isUsageLimitOutcome } from "./error/rate-limit";
 
@@ -90,7 +90,7 @@ export const AUTH_RETRY_STEPS: readonly boolean[] = [false, true];
 export const AUTH_RETRY_MAX_ATTEMPTS = 64;
 
 function isDirectCredentialRotationError(error: unknown): boolean {
-	if (isUsageLimit(error)) return true;
+	if (isUsageLimit(error) || isInvalidatedOAuthTokenError(error)) return true;
 	const status = AIError.status(error);
 	const message = error instanceof Error ? error.message : typeof error === "string" ? error : undefined;
 	return isUsageLimitOutcome(status, message);