@code-yeongyu/senpi-ai 2026.7.31 → 2026.8.3

package/README.md

@@ -874,7 +874,8 @@ for await (const event of s) {
 
 Every `AssistantMessage` includes a `stopReason` field that indicates how the generation ended:
 
-- `"stop"` - Normal completion, the model finished its response
+- `"pending"` - Only present in partial messages when we do not know what the stop reason will be
+- `"stop"` - This is the final message the model will produce this turn
 - `"length"` - Output hit the maximum token limit
 - `"toolUse"` - Model is calling tools and expects tool results
 - `"error"` - An error occurred during generation

package/dist/api/anthropic-messages.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"anthropic-messages.d.ts","sourceRoot":"","sources":["../../src/api/anthropic-messages.ts"],"names":[],"mappings":"AAAA,OAAO,SAAS,MAAM,mBAAmB,CAAC;AAU1C,OAAO,KAAK,EAWX,mBAAmB,EAEnB,cAAc,EACd,aAAa,EAOb,MAAM,aAAa,CAAC;AA+BrB,OAAO,EAAE,kBAAkB,EAAE,MAAM,8BAA8B,CAAC;AA6JlE,MAAM,MAAM,eAAe,GAAG,KAAK,GAAG,QAAQ,GAAG,MAAM,GAAG,OAAO,GAAG,KAAK,CAAC;AAE1E,MAAM,MAAM,wBAAwB,GAAG,YAAY,GAAG,SAAS,CAAC;AAoEhE,MAAM,WAAW,gBAAiB,SAAQ,aAAa;IACtD;;;;;;OAMG;IACH,eAAe,CAAC,EAAE,OAAO,CAAC;IAC1B;;;;OAIG;IACH,oBAAoB,CAAC,EAAE,MAAM,CAAC;IAC9B;;;;;;;;;;;OAWG;IACH,MAAM,CAAC,EAAE,eAAe,CAAC;IACzB;;;;;;;;;;;OAWG;IACH,eAAe,CAAC,EAAE,wBAAwB,CAAC;IAC3C;;;;;OAKG;IACH,mBAAmB,CAAC,EAAE,OAAO,CAAC;IAC9B;;;;OAIG;IACH,UAAU,CAAC,EAAE,MAAM,GAAG,KAAK,GAAG,MAAM,GAAG;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAA;KAAE,CAAC;IACtE;;;;OAIG;IACH,MAAM,CAAC,EAAE,SAAS,CAAC;CACnB;AA8zBD,eAAO,MAAM,MAAM,EAAE,cAAc,CAAC,oBAAoB,EAAE,gBAAgB,CAqYzE,CAAC;AA0FF,eAAO,MAAM,YAAY,EAAE,cAAc,CAAC,oBAAoB,EAAE,mBAAmB,CA2ClF,CAAC"}
+{"version":3,"file":"anthropic-messages.d.ts","sourceRoot":"","sources":["../../src/api/anthropic-messages.ts"],"names":[],"mappings":"AAAA,OAAO,SAAS,MAAM,mBAAmB,CAAC;AAU1C,OAAO,KAAK,EAWX,mBAAmB,EAEnB,cAAc,EACd,aAAa,EAOb,MAAM,aAAa,CAAC;AAiCrB,OAAO,EAAE,kBAAkB,EAAE,MAAM,8BAA8B,CAAC;AA6JlE,MAAM,MAAM,eAAe,GAAG,KAAK,GAAG,QAAQ,GAAG,MAAM,GAAG,OAAO,GAAG,KAAK,CAAC;AAE1E,MAAM,MAAM,wBAAwB,GAAG,YAAY,GAAG,SAAS,CAAC;AAoEhE,MAAM,WAAW,gBAAiB,SAAQ,aAAa;IACtD;;;;;;OAMG;IACH,eAAe,CAAC,EAAE,OAAO,CAAC;IAC1B;;;;OAIG;IACH,oBAAoB,CAAC,EAAE,MAAM,CAAC;IAC9B;;;;;;;;;;;OAWG;IACH,MAAM,CAAC,EAAE,eAAe,CAAC;IACzB;;;;;;;;;;;OAWG;IACH,eAAe,CAAC,EAAE,wBAAwB,CAAC;IAC3C;;;;;OAKG;IACH,mBAAmB,CAAC,EAAE,OAAO,CAAC;IAC9B;;;;OAIG;IACH,UAAU,CAAC,EAAE,MAAM,GAAG,KAAK,GAAG,MAAM,GAAG;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAA;KAAE,CAAC;IACtE;;;;OAIG;IACH,MAAM,CAAC,EAAE,SAAS,CAAC;CACnB;AA4zBD,eAAO,MAAM,MAAM,EAAE,cAAc,CAAC,oBAAoB,EAAE,gBAAgB,CA6ZzE,CAAC;AA0FF,eAAO,MAAM,YAAY,EAAE,cAAc,CAAC,oBAAoB,EAAE,mBAAmB,CA2ClF,CAAC"}

package/dist/api/anthropic-messages.js

@@ -9,10 +9,13 @@ import { parseJsonWithRepair, parseStreamingJson } from "../utils/json-parse.js"
 import { getAnthropicCompat, isAnthropicApiBaseUrl } from "../utils/prompt-cache-ttl.js";
 import { getProviderEnvValue } from "../utils/provider-env.js";
 import { retryProviderRequest } from "../utils/provider-retry.js";
+import { appendRetryAfterMsMarker, extract429RetryAfterMs } from "../utils/retry-hint.js";
 import { sanitizeSurrogates } from "../utils/sanitize-unicode.js";
 import { applyServerFallbackAbort, parseServerFallbackReceipt, parseStickyFallbackReceipt, } from "../utils/server-fallback-receipt.js";
 import { normalizeToolCallId } from "../utils/tool-call-id.js";
 import { isForcedToolChoiceUnsupportedError, omitToolChoiceParam } from "../utils/tool-choice-fallback.js";
+import { demotedToolCallText, demotedToolResultText } from "../utils/unavailable-tool-text.js";
+import { sanitizeAnthropicToolPairs } from "./anthropic-tool-pairs.js";
 import { resolveCloudflareBaseUrl } from "./cloudflare.js";
 import { resolveJsonSchemaStrictSampling } from "./constrained-sampling.js";
 import { buildCopilotDynamicHeaders, hasCopilotVisionInput } from "./github-copilot-headers.js";
@@ -571,6 +574,8 @@ function demoteUnavailableToolReferences(params) {
     if (demotedCallNames.size === 0 && danglingReferenceNames.size === 0)
         return params;
     let changed = false;
+    const availableToolNames = [...definedNames];
+    const seenDemotedCallNames = new Set();
     const rewrittenMessages = [];
     for (const message of messages) {
         if (!Array.isArray(message.content)) {
@@ -584,7 +589,12 @@ function demoteUnavailableToolReferences(params) {
                 const demotedName = demotedCallNames.get(block.id);
                 if (demotedName !== undefined) {
                     messageChanged = true;
-                    content.push({ type: "text", text: demotedToolCallText(demotedName, block.input) });
+                    const firstOccurrence = !seenDemotedCallNames.has(demotedName);
+                    seenDemotedCallNames.add(demotedName);
+                    content.push({
+                        type: "text",
+                        text: demotedToolCallText(demotedName, availableToolNames, firstOccurrence),
+                    });
                     continue;
                 }
             }
@@ -592,7 +602,7 @@ function demoteUnavailableToolReferences(params) {
                 const demotedName = demotedCallNames.get(block.tool_use_id);
                 if (demotedName !== undefined) {
                     messageChanged = true;
-                    content.push({ type: "text", text: demotedToolResultText(demotedName, block.content) });
+                    content.push({ type: "text", text: demotedToolResultText(demotedName, toolResultText(block.content)) });
                     continue;
                 }
                 if (danglingReferenceNames.size > 0 && Array.isArray(block.content)) {
@@ -648,19 +658,6 @@ function collectToolReferenceNames(value, names) {
     for (const nested of Object.values(value))
         collectToolReferenceNames(nested, names);
 }
-function demotedToolCallText(name, input) {
-    let serializedInput;
-    try {
-        serializedInput = JSON.stringify(input ?? {});
-    }
-    catch {
-        serializedInput = "{}";
-    }
-    return `[Called tool "${name}" (no longer available in this session) with input: ${serializedInput}]`;
-}
-function demotedToolResultText(name, content) {
-    return `[Result of unavailable tool "${name}": ${toolResultText(content)}]`;
-}
 function toolResultText(content) {
     if (typeof content === "string")
         return content;
@@ -840,7 +837,12 @@ async function* iterateAnthropicEvents(response, signal) {
     let sawMessageEnd = false;
     for await (const sse of iterateSseMessages(response.body, signal)) {
         if (sse.event === "error") {
-            throw new Error(sse.data);
+            let errorText = sse.data;
+            const hintMs = extract429RetryAfterMs({ bodyText: sse.data });
+            if (hintMs !== undefined) {
+                errorText = appendRetryAfterMsMarker(errorText, hintMs);
+            }
+            throw new Error(errorText);
         }
         if (!ANTHROPIC_MESSAGE_EVENTS.has(sse.event ?? "")) {
             continue;
@@ -881,7 +883,7 @@ export const stream = (model, context, options) => {
                 totalTokens: 0,
                 cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, total: 0 },
             },
-            stopReason: "stop",
+            stopReason: "pending",
             timestamp: Date.now(),
         };
         const serverFallbackAbort = new AbortController();
@@ -911,7 +913,7 @@ export const stream = (model, context, options) => {
                 }
                 const cacheRetention = resolveCacheRetention(options?.cacheRetention ?? model.cacheRetention, options?.env, "long");
                 const cacheSessionId = cacheRetention === "none" ? undefined : options?.sessionId;
-                const created = createClient(model, apiKey, options?.interleavedThinking ?? true, shouldUseFineGrainedToolStreamingBeta(model, context), optionsHeaders, copilotDynamicHeaders, cacheSessionId, options?.env);
+                const created = createClient(model, apiKey, options?.interleavedThinking ?? true, shouldUseFineGrainedToolStreamingBeta(model, context), optionsHeaders, options?.fetch, copilotDynamicHeaders, cacheSessionId, options?.env);
                 client = created.client;
                 isOAuth = created.isOAuthToken;
             }
@@ -927,7 +929,7 @@ export const stream = (model, context, options) => {
                 }
                 params = sanitizeAdaptiveThinkingPayload(model, params, options);
                 params = sanitizeUnsupportedNativeTools(model, params);
-                params = demoteUnavailableToolReferences(params);
+                params = sanitizeAnthropicToolPairs(demoteUnavailableToolReferences(params));
                 const payloadRequestMetadata = extractPayloadRequestMetadata(params);
                 params = payloadRequestMetadata.params;
                 const requestOptions = {
@@ -1007,7 +1009,7 @@ export const stream = (model, context, options) => {
                     if (event.content_block.type === "text") {
                         const block = {
                             type: "text",
-                            text: "",
+                            text: event.content_block.text ?? "",
                             index: event.index,
                         };
                         output.content.push(block);
@@ -1016,8 +1018,8 @@ export const stream = (model, context, options) => {
                     else if (event.content_block.type === "thinking") {
                         const block = {
                             type: "thinking",
-                            thinking: "",
-                            thinkingSignature: "",
+                            thinking: event.content_block.thinking ?? "",
+                            thinkingSignature: event.content_block.signature ?? "",
                             index: event.index,
                         };
                         output.content.push(block);
@@ -1159,6 +1161,7 @@ export const stream = (model, context, options) => {
                 }
                 else if (event.type === "message_delta") {
                     if (event.delta.stop_reason) {
+                        output.rawStopReason = event.delta.stop_reason;
                         const stopReasonResult = mapStopReason(event.delta.stop_reason, event.delta.stop_details);
                         output.stopReason = stopReasonResult.stopReason;
                         if (stopReasonResult.errorMessage) {
@@ -1204,6 +1207,9 @@ export const stream = (model, context, options) => {
             if (serverFallbackReceipt !== undefined) {
                 applyServerFallbackAbort(output, serverFallbackReceipt);
             }
+            if (output.stopReason === "pending") {
+                throw new Error("Anthropic stream ended without a stop reason");
+            }
             if (output.stopReason === "aborted" || output.stopReason === "error") {
                 throw new Error(output.errorMessage || "An unknown error occurred");
             }
@@ -1226,7 +1232,24 @@ export const stream = (model, context, options) => {
                 delete block.partialJson;
             }
             output.stopReason = options?.signal?.aborted ? "aborted" : "error";
-            output.errorMessage = error instanceof Error ? error.message : JSON.stringify(error);
+            let errorMessage = error instanceof Error ? error.message : JSON.stringify(error);
+            // When a 429 SDK error carries response headers, extract the server's
+            // retry-after hint and append the canonical marker so the orchestration
+            // layer can recover the full delay.
+            if (error instanceof Error && error.status === 429) {
+                const sdkHeaders = error.headers;
+                if (sdkHeaders instanceof Headers) {
+                    const hintMs = extract429RetryAfterMs({
+                        status: 429,
+                        headers: sdkHeaders,
+                        bodyText: errorMessage,
+                    });
+                    if (hintMs !== undefined) {
+                        errorMessage = appendRetryAfterMsMarker(errorMessage, hintMs);
+                    }
+                }
+            }
+            output.errorMessage = errorMessage;
             stream.push({ type: "error", reason: output.stopReason, error: output });
             stream.end();
         }
@@ -1336,7 +1359,7 @@ export const streamSimple = (model, context, options) => {
 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, fetch, dynamicHeaders, sessionId, env) {
     // Adaptive thinking models have interleaved thinking built in, so skip the beta header.
     const needsInterleavedBeta = interleavedThinking && !supportsAdaptiveThinking(model);
     const betaFeatures = [];
@@ -1352,6 +1375,7 @@ function createClient(model, apiKey, interleavedThinking, useFineGrainedToolStre
             authToken: null,
             baseURL: resolveCloudflareBaseUrl(model, env),
             dangerouslyAllowBrowser: true,
+            fetch,
             defaultHeaders: sanitizeAdaptiveThinkingHeaders(model, mergeHeaders({
                 accept: "application/json",
                 "anthropic-dangerous-direct-browser-access": "true",
@@ -1370,6 +1394,7 @@ function createClient(model, apiKey, interleavedThinking, useFineGrainedToolStre
             authToken: apiKey,
             baseURL: model.baseUrl,
             dangerouslyAllowBrowser: true,
+            fetch,
             defaultHeaders: sanitizeAdaptiveThinkingHeaders(model, mergeHeaders({
                 accept: "application/json",
                 "anthropic-dangerous-direct-browser-access": "true",
@@ -1385,6 +1410,7 @@ function createClient(model, apiKey, interleavedThinking, useFineGrainedToolStre
             authToken: apiKey,
             baseURL: model.baseUrl,
             dangerouslyAllowBrowser: true,
+            fetch,
             defaultHeaders: sanitizeAdaptiveThinkingHeaders(model, mergeHeaders({
                 accept: "application/json",
                 "anthropic-dangerous-direct-browser-access": "true",
@@ -1402,6 +1428,7 @@ function createClient(model, apiKey, interleavedThinking, useFineGrainedToolStre
         authToken: null,
         baseURL: model.baseUrl,
         dangerouslyAllowBrowser: true,
+        fetch,
         defaultHeaders: sanitizeAdaptiveThinkingHeaders(model, mergeHeaders({
             accept: "application/json",
             "anthropic-dangerous-direct-browser-access": "true",
@@ -1847,7 +1874,11 @@ function mapStopReason(reason, stopDetails) {
         case "stop_sequence":
             return { stopReason: "stop" }; // We don't supply stop sequences, so this should never happen
         case "sensitive": // Content flagged by safety filters (not yet in SDK types)
-            return { stopReason: "error", stopDetails: { type: "sensitive" } };
+            return {
+                stopReason: "error",
+                errorMessage: "Provider stopped with: sensitive",
+                stopDetails: { type: "sensitive" },
+            };
         default:
             // Handle unknown stop reasons gracefully (API may add new values)
             throw new Error(`Unhandled stop reason: ${reason}`);