@psnext/slingcli 2.5.20260623-1 → 2.5.20260629-1

package/node_modules/@earendil-works/pi-agent-core/dist/harness/compaction/compaction.js

@@ -65,13 +65,16 @@ export function calculateContextTokens(usage) {
 function getAssistantUsage(msg) {
     if (msg.role === "assistant" && "usage" in msg) {
         const assistantMsg = msg;
-        if (assistantMsg.stopReason !== "aborted" && assistantMsg.stopReason !== "error" && assistantMsg.usage) {
+        if (assistantMsg.stopReason !== "aborted" &&
+            assistantMsg.stopReason !== "error" &&
+            assistantMsg.usage &&
+            calculateContextTokens(assistantMsg.usage) > 0) {
             return assistantMsg.usage;
         }
     }
     return undefined;
 }
-/** Return usage from the last successful assistant message in session entries. */
+/** Return usage from the last valid assistant message in session entries. */
 export function getLastAssistantUsage(entries) {
     for (let i = entries.length - 1; i >= 0; i--) {
         const entry = entries[i];

package/node_modules/@earendil-works/pi-agent-core/dist/harness/session/session.js

@@ -180,12 +180,13 @@ export class Session {
         });
     }
     async appendSessionName(name) {
+        const sanitizedName = name.replace(/[\r\n]+/g, " ").trim();
         return this.appendTypedEntry({
             type: "session_info",
             id: await this.storage.createEntryId(),
             parentId: await this.storage.getLeafId(),
             timestamp: new Date().toISOString(),
-            name: name.trim(),
+            name: sanitizedName,
         });
     }
     async moveTo(entryId, summary) {

package/node_modules/@earendil-works/pi-agent-core/dist/harness/utils/shell-output.js

@@ -1,4 +1,4 @@
-import { ExecutionError, err, ok, toError, } from "../types.js";
+import { ExecutionError, err, ok, toError } from "../types.js";
 import { DEFAULT_MAX_BYTES, truncateTail } from "./truncate.js";
 function toExecutionError(error) {
     if (error instanceof ExecutionError)

package/node_modules/@earendil-works/pi-agent-core/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@earendil-works/pi-agent-core",
-	"version": "0.79.10",
+	"version": "0.80.2",
 	"description": "General-purpose agent with transport abstraction, state management, and attachment support",
 	"type": "module",
 	"main": "./dist/index.js",
@@ -29,7 +29,7 @@
 		"prepublishOnly": "npm run clean && npm run build"
 	},
 	"dependencies": {
-		"@earendil-works/pi-ai": "^0.79.10",
+		"@earendil-works/pi-ai": "^0.80.2",
 		"ignore": "7.0.5",
 		"typebox": "1.1.38",
 		"yaml": "2.9.0"

package/node_modules/@earendil-works/pi-ai/dist/api/anthropic-messages.js

@@ -5,9 +5,8 @@ import { headersToRecord } from "../utils/headers.js";
 import { parseJsonWithRepair, parseStreamingJson } from "../utils/json-parse.js";
 import { getProviderEnvValue } from "../utils/provider-env.js";
 import { sanitizeSurrogates } from "../utils/sanitize-unicode.js";
-import { resolveCloudflareBaseUrl } from "./cloudflare.js";
 import { buildCopilotDynamicHeaders, hasCopilotVisionInput } from "./github-copilot-headers.js";
-import { adjustMaxTokensForThinking, buildBaseOptions } from "./simple-options.js";
+import { adjustMaxTokensForThinking, buildBaseOptions, clampMaxTokensToContext } from "./simple-options.js";
 import { transformMessages } from "./transform-messages.js";
 /**
  * Resolve cache retention preference.
@@ -108,14 +107,11 @@ function convertContentBlocks(content) {
 const FINE_GRAINED_TOOL_STREAMING_BETA = "fine-grained-tool-streaming-2025-05-14";
 const INTERLEAVED_THINKING_BETA = "interleaved-thinking-2025-05-14";
 function getAnthropicCompat(model) {
-    // Auto-detect session affinity and cache control support from provider
-    const isFireworks = model.provider === "fireworks";
-    const isCloudflareAiGatewayAnthropic = model.provider === "cloudflare-ai-gateway" && model.baseUrl.includes("anthropic");
     return {
-        supportsEagerToolInputStreaming: model.compat?.supportsEagerToolInputStreaming ?? !isFireworks,
-        supportsLongCacheRetention: model.compat?.supportsLongCacheRetention ?? !isFireworks,
-        sendSessionAffinityHeaders: model.compat?.sendSessionAffinityHeaders ?? !!(isFireworks || isCloudflareAiGatewayAnthropic),
-        supportsCacheControlOnTools: model.compat?.supportsCacheControlOnTools ?? !isFireworks,
+        supportsEagerToolInputStreaming: model.compat?.supportsEagerToolInputStreaming ?? true,
+        supportsLongCacheRetention: model.compat?.supportsLongCacheRetention ?? true,
+        sendSessionAffinityHeaders: model.compat?.sendSessionAffinityHeaders ?? false,
+        supportsCacheControlOnTools: model.compat?.supportsCacheControlOnTools ?? true,
         supportsTemperature: model.compat?.supportsTemperature ?? true,
         allowEmptySignature: model.compat?.allowEmptySignature ?? false,
     };
@@ -129,6 +125,26 @@ function mergeHeaders(...headerSources) {
     }
     return merged;
 }
+function hasHeader(headers, name) {
+    if (!headers)
+        return false;
+    const expected = name.toLowerCase();
+    for (const [key, value] of Object.entries(headers)) {
+        if (key.toLowerCase() === expected && value !== null && value.trim().length > 0)
+            return true;
+    }
+    return false;
+}
+function assertRequestAuth(provider, apiKey, headers) {
+    if (apiKey)
+        return;
+    if (hasHeader(headers, "authorization") ||
+        hasHeader(headers, "x-api-key") ||
+        hasHeader(headers, "cf-aig-authorization")) {
+        return;
+    }
+    throw new Error(`No API key for provider: ${provider}`);
+}
 const ANTHROPIC_MESSAGE_EVENTS = new Set([
     "message_start",
     "message_delta",
@@ -309,9 +325,7 @@ export const stream = (model, context, options) => {
             }
             else {
                 const apiKey = options?.apiKey;
-                if (!apiKey) {
-                    throw new Error(`No API key for provider: ${model.provider}`);
-                }
+                assertRequestAuth(model.provider, apiKey, options?.headers);
                 let copilotDynamicHeaders;
                 if (model.provider === "github-copilot") {
                     const hasImages = hasCopilotVisionInput(context.messages);
@@ -322,7 +336,7 @@ export const stream = (model, context, options) => {
                 }
                 const cacheRetention = resolveCacheRetention(options?.cacheRetention, options?.env);
                 const cacheSessionId = cacheRetention === "none" ? undefined : options?.sessionId;
-                const created = createClient(model, apiKey, options?.interleavedThinking ?? true, shouldUseFineGrainedToolStreamingBeta(model, context), options?.headers, copilotDynamicHeaders, cacheSessionId, options?.env);
+                const created = createClient(model, apiKey, options?.interleavedThinking ?? true, shouldUseFineGrainedToolStreamingBeta(model, context), options?.headers, copilotDynamicHeaders, cacheSessionId);
                 client = created.client;
                 isOAuth = created.isOAuthToken;
             }
@@ -508,6 +522,14 @@ export const stream = (model, context, options) => {
                     if (event.usage.cache_creation_input_tokens != null) {
                         output.usage.cacheWrite = event.usage.cache_creation_input_tokens;
                     }
+                    // Anthropic reports reasoning tokens in `output_tokens_details.thinking_tokens` on the
+                    // final message_delta usage (a subset of output_tokens). SDK 0.91.1 omits the field from
+                    // its Usage type, so read it through a narrow cast. Verified against the live API.
+                    const thinkingTokens = event.usage
+                        .output_tokens_details?.thinking_tokens;
+                    if (thinkingTokens != null) {
+                        output.usage.reasoning = thinkingTokens;
+                    }
                     // Anthropic doesn't provide total_tokens, compute from components
                     output.usage.totalTokens =
                         output.usage.input + output.usage.output + output.usage.cacheRead + output.usage.cacheWrite;
@@ -558,11 +580,8 @@ function mapThinkingLevelToEffort(model, level) {
     }
 }
 export const streamSimple = (model, context, options) => {
-    const apiKey = options?.apiKey;
-    if (!apiKey) {
-        throw new Error(`No API key for provider: ${model.provider}`);
-    }
-    const base = buildBaseOptions(model, options, apiKey);
+    assertRequestAuth(model.provider, options?.apiKey, options?.headers);
+    const base = buildBaseOptions(model, context, options, options?.apiKey);
     if (!options?.reasoning) {
         return stream(model, context, { ...base, thinkingEnabled: false });
     }
@@ -579,17 +598,18 @@ export const streamSimple = (model, context, options) => {
     // Undefined means the caller did not request an output cap; let the helper use the model cap.
     // Do not coerce to 0 here, or the thinking budget would become the entire max_tokens value.
     const adjusted = adjustMaxTokensForThinking(base.maxTokens, model.maxTokens, options.reasoning, options.thinkingBudgets);
+    const maxTokens = clampMaxTokensToContext(model, context, adjusted.maxTokens);
     return stream(model, context, {
         ...base,
-        maxTokens: adjusted.maxTokens,
+        maxTokens,
         thinkingEnabled: true,
-        thinkingBudgetTokens: adjusted.thinkingBudget,
+        thinkingBudgetTokens: Math.min(adjusted.thinkingBudget, Math.max(0, maxTokens - 1024)),
     });
 };
 function isOAuthToken(apiKey) {
     return apiKey.includes("sk-ant-oat");
 }
-function createClient(model, apiKey, interleavedThinking, useFineGrainedToolStreamingBeta, optionsHeaders, dynamicHeaders, sessionId, env) {
+function createClient(model, apiKey, interleavedThinking, useFineGrainedToolStreamingBeta, optionsHeaders, dynamicHeaders, sessionId) {
     // Adaptive thinking models have interleaved thinking built in, so skip the beta header.
     const needsInterleavedBeta = interleavedThinking && model.compat?.forceAdaptiveThinking !== true;
     const betaFeatures = [];
@@ -599,28 +619,11 @@ function createClient(model, apiKey, interleavedThinking, useFineGrainedToolStre
     if (needsInterleavedBeta) {
         betaFeatures.push(INTERLEAVED_THINKING_BETA);
     }
-    if (model.provider === "cloudflare-ai-gateway") {
-        const client = new Anthropic({
-            apiKey: null,
-            authToken: null,
-            baseURL: resolveCloudflareBaseUrl(model, env),
-            dangerouslyAllowBrowser: true,
-            defaultHeaders: mergeHeaders({
-                accept: "application/json",
-                "anthropic-dangerous-direct-browser-access": "true",
-                "cf-aig-authorization": `Bearer ${apiKey}`,
-                "x-api-key": null,
-                Authorization: null,
-                ...(betaFeatures.length > 0 ? { "anthropic-beta": betaFeatures.join(",") } : {}),
-            }, model.headers, optionsHeaders),
-        });
-        return { client, isOAuthToken: false };
-    }
     // Copilot: Bearer auth, selective betas.
     if (model.provider === "github-copilot") {
         const client = new Anthropic({
             apiKey: null,
-            authToken: apiKey,
+            authToken: apiKey ?? null,
             baseURL: model.baseUrl,
             dangerouslyAllowBrowser: true,
             defaultHeaders: mergeHeaders({
@@ -632,7 +635,7 @@ function createClient(model, apiKey, interleavedThinking, useFineGrainedToolStre
         return { client, isOAuthToken: false };
     }
     // OAuth: Bearer auth, Claude Code identity headers
-    if (isOAuthToken(apiKey)) {
+    if (apiKey && isOAuthToken(apiKey)) {
         const client = new Anthropic({
             apiKey: null,
             authToken: apiKey,
@@ -648,18 +651,19 @@ function createClient(model, apiKey, interleavedThinking, useFineGrainedToolStre
         });
         return { client, isOAuthToken: true };
     }
-    // API key auth
+    // API key or header-owned auth.
     const sessionAffinityHeaders = sessionId && getAnthropicCompat(model).sendSessionAffinityHeaders ? { "x-session-affinity": sessionId } : {};
+    const defaultHeaders = mergeHeaders({
+        accept: "application/json",
+        "anthropic-dangerous-direct-browser-access": "true",
+        ...(betaFeatures.length > 0 ? { "anthropic-beta": betaFeatures.join(",") } : {}),
+    }, sessionAffinityHeaders, model.headers, optionsHeaders);
     const client = new Anthropic({
-        apiKey,
+        apiKey: apiKey ?? null,
         authToken: null,
         baseURL: model.baseUrl,
         dangerouslyAllowBrowser: true,
-        defaultHeaders: mergeHeaders({
-            accept: "application/json",
-            "anthropic-dangerous-direct-browser-access": "true",
-            ...(betaFeatures.length > 0 ? { "anthropic-beta": betaFeatures.join(",") } : {}),
-        }, sessionAffinityHeaders, model.headers, optionsHeaders),
+        defaultHeaders,
     });
     return { client, isOAuthToken: false };
 }

package/node_modules/@earendil-works/pi-ai/dist/api/azure-openai-responses.js

@@ -120,7 +120,7 @@ export const streamSimple = (model, context, options) => {
     if (!apiKey) {
         throw new Error(`No API key for provider: ${model.provider}`);
     }
-    const base = buildBaseOptions(model, options, apiKey);
+    const base = buildBaseOptions(model, context, options, apiKey);
     const clampedReasoning = options?.reasoning ? clampThinkingLevel(model, options.reasoning) : undefined;
     const reasoningEffort = clampedReasoning === "off" ? undefined : clampedReasoning;
     return stream(model, context, {
@@ -137,11 +137,17 @@ function normalizeAzureBaseUrl(baseUrl) {
     catch {
         throw new Error(`Invalid Azure OpenAI base URL: ${baseUrl}`);
     }
-    const isAzureHost = url.hostname.endsWith(".openai.azure.com") || url.hostname.endsWith(".cognitiveservices.azure.com");
+    const isAzureHost = url.hostname.endsWith(".openai.azure.com") ||
+        url.hostname.endsWith(".cognitiveservices.azure.com") ||
+        url.hostname.endsWith(".ai.azure.com");
     const normalizedPath = url.pathname.replace(/\/+$/, "");
     // Ensure Azure hosts have /openai/v1 as base path so the AzureOpenAI SDK
     // can append /deployments/<model>/... and ?api-version=v1 correctly.
-    if (isAzureHost && (normalizedPath === "" || normalizedPath === "/" || normalizedPath === "/openai")) {
+    if (isAzureHost &&
+        (normalizedPath === "" ||
+            normalizedPath === "/" ||
+            normalizedPath === "/openai" ||
+            normalizedPath === "/openai/v1/responses")) {
         url.pathname = "/openai/v1";
         url.search = "";
     }

package/node_modules/@earendil-works/pi-ai/dist/api/bedrock-converse-stream.js

@@ -4,11 +4,12 @@ import { HttpProxyAgent } from "http-proxy-agent";
 import { HttpsProxyAgent } from "https-proxy-agent";
 import { calculateCost } from "../models.js";
 import { AssistantMessageEventStream } from "../utils/event-stream.js";
+import { providerHeadersToRecord } from "../utils/headers.js";
 import { parseStreamingJson } from "../utils/json-parse.js";
 import { resolveHttpProxyUrlForTarget } from "../utils/node-http-proxy.js";
 import { getProviderEnvValue } from "../utils/provider-env.js";
 import { sanitizeSurrogates } from "../utils/sanitize-unicode.js";
-import { adjustMaxTokensForThinking, buildBaseOptions, clampReasoning } from "./simple-options.js";
+import { adjustMaxTokensForThinking, buildBaseOptions, clampMaxTokensToContext, clampReasoning, } from "./simple-options.js";
 import { transformMessages } from "./transform-messages.js";
 const EMPTY_TEXT_PLACEHOLDER = "<empty>";
 export const stream = (model, context, options = {}) => {
@@ -105,8 +106,9 @@ export const stream = (model, context, options = {}) => {
         }
         try {
             const client = new BedrockRuntimeClient(config);
-            if (options.headers && Object.keys(options.headers).length > 0) {
-                addCustomHeadersMiddleware(client, options.headers);
+            const customHeaders = providerHeadersToRecord(options.headers);
+            if (customHeaders) {
+                addCustomHeadersMiddleware(client, customHeaders);
             }
             const cacheRetention = resolveCacheRetention(options.cacheRetention, options.env);
             const inferenceMaxTokens = options.maxTokens ?? (isAnthropicClaudeModel(model) ? model.maxTokens : undefined);
@@ -267,7 +269,7 @@ function addCustomHeadersMiddleware(client, headers) {
     client.middlewareStack.add(middleware, { step: "build", name: "pi-ai-custom-headers", priority: "low" });
 }
 export const streamSimple = (model, context, options) => {
-    const base = buildBaseOptions(model, options, undefined);
+    const base = buildBaseOptions(model, context, options, undefined);
     if (!options?.reasoning) {
         return stream(model, context, { ...base, reasoning: undefined });
     }
@@ -282,13 +284,14 @@ export const streamSimple = (model, context, options) => {
         // Undefined means the caller did not request an output cap; let the helper use the model cap.
         // Do not coerce to 0 here, or the thinking budget would become the entire maxTokens value.
         const adjusted = adjustMaxTokensForThinking(base.maxTokens, model.maxTokens, options.reasoning, options.thinkingBudgets);
+        const maxTokens = clampMaxTokensToContext(model, context, adjusted.maxTokens);
         return stream(model, context, {
             ...base,
-            maxTokens: adjusted.maxTokens,
+            maxTokens,
             reasoning: options.reasoning,
             thinkingBudgets: {
                 ...(options.thinkingBudgets || {}),
-                [clampReasoning(options.reasoning)]: adjusted.thinkingBudget,
+                [clampReasoning(options.reasoning)]: Math.min(adjusted.thinkingBudget, Math.max(0, maxTokens - 1024)),
             },
         });
     }

package/node_modules/@earendil-works/pi-ai/dist/api/cloudflare.js

@@ -1,4 +1,3 @@
-import { getProviderEnvValue } from "../utils/provider-env.js";
 /** Workers AI direct endpoint. */
 export const CLOUDFLARE_WORKERS_AI_BASE_URL = "https://api.cloudflare.com/client/v4/accounts/{CLOUDFLARE_ACCOUNT_ID}/ai/v1";
 /** AI Gateway Unified API. https://developers.cloudflare.com/ai-gateway/usage/unified-api/ */
@@ -7,21 +6,4 @@ export const CLOUDFLARE_AI_GATEWAY_COMPAT_BASE_URL = "https://gateway.ai.cloudfl
 export const CLOUDFLARE_AI_GATEWAY_OPENAI_BASE_URL = "https://gateway.ai.cloudflare.com/v1/{CLOUDFLARE_ACCOUNT_ID}/{CLOUDFLARE_GATEWAY_ID}/openai";
 /** AI Gateway → Anthropic passthrough. */
 export const CLOUDFLARE_AI_GATEWAY_ANTHROPIC_BASE_URL = "https://gateway.ai.cloudflare.com/v1/{CLOUDFLARE_ACCOUNT_ID}/{CLOUDFLARE_GATEWAY_ID}/anthropic";
-export function isCloudflareProvider(provider) {
-    return provider === "cloudflare-workers-ai" || provider === "cloudflare-ai-gateway";
-}
-/** Substitute `{VAR}` placeholders in a Cloudflare baseUrl from provider env or process.env. */
-export function resolveCloudflareBaseUrl(model, env) {
-    const url = model.baseUrl;
-    if (!url.includes("{"))
-        return url;
-    const baseUrl = url.replace(/\{([A-Z_][A-Z0-9_]*)\}/g, (_match, name) => {
-        const value = getProviderEnvValue(name, env);
-        if (!value) {
-            throw new Error(`${name} is required for provider ${model.provider} but is not set.`);
-        }
-        return value;
-    });
-    return baseUrl;
-}
 //# sourceMappingURL=cloudflare.js.map

package/node_modules/@earendil-works/pi-ai/dist/api/google-generative-ai.js

@@ -1,6 +1,7 @@
 import { GoogleGenAI, } from "@google/genai";
 import { calculateCost, clampThinkingLevel } from "../models.js";
 import { AssistantMessageEventStream } from "../utils/event-stream.js";
+import { providerHeadersToRecord } from "../utils/headers.js";
 import { sanitizeSurrogates } from "../utils/sanitize-unicode.js";
 import { convertMessages, convertTools, isThinkingPart, mapStopReason, mapToolChoice, retainThoughtSignature, } from "./google-shared.js";
 import { buildBaseOptions } from "./simple-options.js";
@@ -161,6 +162,7 @@ export const stream = (model, context, options) => {
                         output: (chunk.usageMetadata.candidatesTokenCount || 0) + (chunk.usageMetadata.thoughtsTokenCount || 0),
                         cacheRead: chunk.usageMetadata.cachedContentTokenCount || 0,
                         cacheWrite: 0,
+                        reasoning: chunk.usageMetadata.thoughtsTokenCount || 0,
                         totalTokens: chunk.usageMetadata.totalTokenCount || 0,
                         cost: {
                             input: 0,
@@ -220,7 +222,7 @@ export const streamSimple = (model, context, options) => {
     if (!apiKey) {
         throw new Error(`No API key for provider: ${model.provider}`);
     }
-    const base = buildBaseOptions(model, options, apiKey);
+    const base = buildBaseOptions(model, context, options, apiKey);
     if (!options?.reasoning) {
         return stream(model, context, { ...base, thinking: { enabled: false } });
     }
@@ -250,8 +252,9 @@ function createClient(model, apiKey, optionsHeaders) {
         httpOptions.baseUrl = model.baseUrl;
         httpOptions.apiVersion = ""; // baseUrl already includes version path, don't append
     }
-    if (model.headers || optionsHeaders) {
-        httpOptions.headers = { ...model.headers, ...optionsHeaders };
+    const headers = providerHeadersToRecord({ ...model.headers, ...optionsHeaders });
+    if (headers) {
+        httpOptions.headers = headers;
     }
     return new GoogleGenAI({
         apiKey,

package/node_modules/@earendil-works/pi-ai/dist/api/google-vertex.js

@@ -1,6 +1,7 @@
 import { GoogleGenAI, ResourceScope, ThinkingLevel, } from "@google/genai";
 import { calculateCost, clampThinkingLevel } from "../models.js";
 import { AssistantMessageEventStream } from "../utils/event-stream.js";
+import { providerHeadersToRecord } from "../utils/headers.js";
 import { getProviderEnvValue } from "../utils/provider-env.js";
 import { sanitizeSurrogates } from "../utils/sanitize-unicode.js";
 import { convertMessages, convertTools, isThinkingPart, mapStopReason, mapToolChoice, retainThoughtSignature, } from "./google-shared.js";
@@ -170,6 +171,7 @@ export const stream = (model, context, options) => {
                         output: (chunk.usageMetadata.candidatesTokenCount || 0) + (chunk.usageMetadata.thoughtsTokenCount || 0),
                         cacheRead: chunk.usageMetadata.cachedContentTokenCount || 0,
                         cacheWrite: 0,
+                        reasoning: chunk.usageMetadata.thoughtsTokenCount || 0,
                         totalTokens: chunk.usageMetadata.totalTokenCount || 0,
                         cost: {
                             input: 0,
@@ -225,7 +227,7 @@ export const stream = (model, context, options) => {
     return stream;
 };
 export const streamSimple = (model, context, options) => {
-    const base = buildBaseOptions(model, options, undefined);
+    const base = buildBaseOptions(model, context, options, undefined);
     if (!options?.reasoning) {
         return stream(model, context, {
             ...base,
@@ -281,8 +283,9 @@ function buildHttpOptions(model, optionsHeaders) {
             httpOptions.apiVersion = "";
         }
     }
-    if (model.headers || optionsHeaders) {
-        httpOptions.headers = { ...model.headers, ...optionsHeaders };
+    const headers = providerHeadersToRecord({ ...model.headers, ...optionsHeaders });
+    if (headers) {
+        httpOptions.headers = headers;
     }
     return Object.keys(httpOptions).length > 0 ? httpOptions : undefined;
 }

package/node_modules/@earendil-works/pi-ai/dist/api/mistral-conversations.js

@@ -65,7 +65,7 @@ export const streamSimple = (model, context, options) => {
     if (!apiKey) {
         throw new Error(`No API key for provider: ${model.provider}`);
     }
-    const base = buildBaseOptions(model, options, apiKey);
+    const base = buildBaseOptions(model, context, options, apiKey);
     const clampedReasoning = options?.reasoning ? clampThinkingLevel(model, options.reasoning) : undefined;
     const reasoning = clampedReasoning === "off" ? undefined : clampedReasoning;
     const shouldUseReasoning = model.reasoning && reasoning !== undefined;

package/node_modules/@earendil-works/pi-ai/dist/api/openai-codex-responses.js

@@ -33,9 +33,6 @@ const JWT_CLAIM_PATH = "https://api.openai.com/auth";
 const DEFAULT_MAX_RETRIES = 0;
 const BASE_DELAY_MS = 1000;
 const DEFAULT_MAX_RETRY_DELAY_MS = 60_000;
-// Keep a bounded pre-header timeout so zero-event Codex SSE stalls fail instead of
-// leaving callers stuck on "Working..." indefinitely. See #4945.
-const DEFAULT_SSE_HEADER_TIMEOUT_MS = 20_000;
 const DEFAULT_WEBSOCKET_CONNECT_TIMEOUT_MS = 15_000;
 const CODEX_TOOL_CALL_PROVIDERS = new Set(["openai", "openai-codex", "opencode"]);
 const WEBSOCKET_MESSAGE_TOO_BIG_CLOSE_CODE = 1009;
@@ -110,19 +107,6 @@ function normalizeTimeoutMs(value) {
     }
     return Math.floor(value);
 }
-function createSSEHeaderTimeout() {
-    const controller = new AbortController();
-    let error;
-    const timeout = setTimeout(() => {
-        error = new Error(`Codex SSE response headers timed out after ${DEFAULT_SSE_HEADER_TIMEOUT_MS}ms`);
-        controller.abort(error);
-    }, DEFAULT_SSE_HEADER_TIMEOUT_MS);
-    return {
-        signal: controller.signal,
-        clear: () => clearTimeout(timeout),
-        error: () => error,
-    };
-}
 // ============================================================================
 // Main Stream Function
 // ============================================================================
@@ -161,7 +145,7 @@ export const stream = (model, context, options) => {
             const sseHeaders = buildSSEHeaders(model.headers, options?.headers, accountId, apiKey, options?.sessionId);
             const websocketHeaders = buildWebSocketHeaders(model.headers, options?.headers, accountId, apiKey, websocketRequestId);
             const bodyJson = JSON.stringify(body);
-            const idleTimeoutMs = normalizeTimeoutMs(options?.timeoutMs);
+            const httpTimeoutMs = normalizeTimeoutMs(options?.timeoutMs);
             const websocketConnectTimeoutMs = normalizeTimeoutMs(options?.websocketConnectTimeoutMs);
             const transport = options?.transport || "auto";
             const websocketDisabledForSession = transport !== "sse" && isWebSocketSseFallbackActive(options?.sessionId);
@@ -176,7 +160,7 @@ export const stream = (model, context, options) => {
                     try {
                         await processWebSocketStream(resolveCodexWebSocketUrl(model.baseUrl), body, websocketHeaders, output, stream, model, () => {
                             websocketStarted = true;
-                        }, idleTimeoutMs, websocketConnectTimeoutMs, options);
+                        }, httpTimeoutMs, websocketConnectTimeoutMs, options);
                         if (options?.signal?.aborted) {
                             throw new Error("Request was aborted");
                         }
@@ -223,8 +207,8 @@ export const stream = (model, context, options) => {
                     throw new Error("Request was aborted");
                 }
                 try {
-                    const headerTimeout = createSSEHeaderTimeout();
-                    const combinedSignal = combineAbortSignals([options?.signal, headerTimeout.signal]);
+                    const headerTimeoutSignal = httpTimeoutMs !== undefined && httpTimeoutMs > 0 ? AbortSignal.timeout(httpTimeoutMs) : undefined;
+                    const combinedSignal = combineAbortSignals([options?.signal, headerTimeoutSignal]);
                     try {
                         response = await fetch(resolveCodexUrl(model.baseUrl), {
                             method: "POST",
@@ -234,12 +218,13 @@ export const stream = (model, context, options) => {
                         });
                     }
                     catch (error) {
-                        const timeoutError = headerTimeout.error();
-                        throw timeoutError && !options?.signal?.aborted ? timeoutError : error;
+                        if (headerTimeoutSignal?.aborted && !options?.signal?.aborted) {
+                            throw new Error(`Codex SSE response headers timed out after ${httpTimeoutMs}ms`);
+                        }
+                        throw error;
                     }
                     finally {
                         combinedSignal.cleanup();
-                        headerTimeout.clear();
                     }
                     await options?.onResponse?.({ status: response.status, headers: headersToRecord(response.headers) }, model);
                     if (response.ok) {
@@ -312,7 +297,7 @@ export const streamSimple = (model, context, options) => {
     if (!apiKey) {
         throw new Error(`No API key for provider: ${model.provider}`);
     }
-    const base = buildBaseOptions(model, options, apiKey);
+    const base = buildBaseOptions(model, context, options, apiKey);
     const clampedReasoning = options?.reasoning ? clampThinkingLevel(model, options.reasoning) : undefined;
     const reasoningEffort = clampedReasoning === "off" ? undefined : clampedReasoning;
     return stream(model, context, {
@@ -1164,7 +1149,12 @@ function createCodexRequestId() {
 function buildBaseCodexHeaders(initHeaders, additionalHeaders, accountId, token) {
     const headers = new Headers(initHeaders);
     for (const [key, value] of Object.entries(additionalHeaders || {})) {
-        headers.set(key, value);
+        if (value === null) {
+            headers.delete(key);
+        }
+        else {
+            headers.set(key, value);
+        }
     }
     headers.set("Authorization", `Bearer ${token}`);
     headers.set("chatgpt-account-id", accountId);