diff --git a/packages/agent/CHANGELOG.md b/packages/agent/CHANGELOG.md
index 687e1cb1f..9d0472231 100644
--- a/packages/agent/CHANGELOG.md
+++ b/packages/agent/CHANGELOG.md
@@ -1,10 +1,19 @@
 # Changelog
 
 ## [Unreleased]
+
 ### Removed
 
 - Removed the `maxToolCallsPerTurn` option from `AgentOptions` and `AgentLoopConfig`, so assistant turns are no longer capped after a configured number of completed tool calls
 
+### Fixed
+
+- Fixed tool result parsing to mark assistant tool outputs with unsupported content block shapes as errors and include a diagnostic text block
+- Fixed GPT-5 Harmony leakage handling by recovering valid leaked tool calls when possible and discarding leaked partial assistant output before retrying
+- Fixed tool-call cancellation handling so aborted tools are marked aborted with an explicit reason and do not report generic errors
+- Fixed tool-call completion so assistant messages on abort keep only completed tool-call blocks and continue processing tool calls when a length stop still included results
+- Fixed runs that stopped with reason `length` after returning tool results so execution continues to handle additional tool calls
+
 ## [15.10.3] - 2026-06-08
 
 ### Added
diff --git a/packages/agent/src/agent-loop.ts b/packages/agent/src/agent-loop.ts
index 052b72afb..fc859bb0c 100644
--- a/packages/agent/src/agent-loop.ts
+++ b/packages/agent/src/agent-loop.ts
@@ -20,6 +20,7 @@ import {
 	createHarmonyAuditEvent,
 	detectHarmonyLeakInAssistantMessage,
 	extractHarmonyRemoved,
+	recoverHarmonyToolCall,
 	type HarmonyDetection,
 	type HarmonyRecoveredToolCall,
 	isHarmonyLeakMitigationTarget,
@@ -32,7 +33,6 @@ import {
 	finishChatSpan,
 	finishExecuteToolSpan,
 	finishInvokeAgentSpan,
-	fireOnRunEnd,
 	PiGenAIAttr,
 	recordSkippedTool,
 	resolveTelemetry,
@@ -68,6 +68,76 @@ class HarmonyLeakInterruption extends Error {
 	}
 }
 
+type AssistantContentBlock = AssistantMessage["content"][number];
+type AssistantToolCallBlock = Extract<AssistantContentBlock, { type: "toolCall" }>;
+type CloneableRecord = Record<string, unknown>;
+
+function cloneUnknown(value: unknown): unknown {
+	if (Array.isArray(value)) return value.map(cloneUnknown);
+	if (!value || typeof value !== "object") return value;
+	const source = value as CloneableRecord;
+	const out: CloneableRecord = {};
+	for (const [key, child] of Object.entries(source)) {
+		out[key] = cloneUnknown(child);
+	}
+	return out;
+}
+
+function cloneToolArguments(args: AssistantToolCallBlock["arguments"]): AssistantToolCallBlock["arguments"] {
+	return cloneUnknown(args) as AssistantToolCallBlock["arguments"];
+}
+
+function snapshotAssistantContentBlock(block: AssistantContentBlock): AssistantContentBlock {
+	switch (block.type) {
+		case "text":
+			return { ...block };
+		case "thinking":
+			return { ...block };
+		case "redactedThinking":
+			return { ...block };
+		case "toolCall":
+			return { ...block, arguments: cloneToolArguments(block.arguments) };
+	}
+}
+
+function snapshotAssistantMessage(message: AssistantMessage): AssistantMessage {
+	return {
+		...message,
+		content: message.content.map(snapshotAssistantContentBlock),
+		usage: {
+			...message.usage,
+			cost: { ...message.usage.cost },
+		},
+		disabledFeatures: message.disabledFeatures ? [...message.disabledFeatures] : undefined,
+	};
+}
+
+function snapshotAssistantMessageEvent(event: AssistantMessageEvent): AssistantMessageEvent {
+	switch (event.type) {
+		case "start":
+			return { ...event, partial: snapshotAssistantMessage(event.partial) };
+		case "text_start":
+		case "text_delta":
+		case "text_end":
+		case "thinking_start":
+		case "thinking_delta":
+		case "thinking_end":
+		case "toolcall_start":
+		case "toolcall_delta":
+			return { ...event, partial: snapshotAssistantMessage(event.partial) };
+		case "toolcall_end":
+			return {
+				...event,
+				toolCall: snapshotAssistantContentBlock(event.toolCall) as AssistantToolCallBlock,
+				partial: snapshotAssistantMessage(event.partial),
+			};
+		case "done":
+			return { ...event, message: snapshotAssistantMessage(event.message) };
+		case "error":
+			return { ...event, error: snapshotAssistantMessage(event.error) };
+	}
+}
+
 /**
  * Normalize a value coming back from `tool.execute()` (or its streaming partial-update callback)
  * into a structurally valid {@link AgentToolResult}.
@@ -77,7 +147,7 @@ class HarmonyLeakInterruption extends Error {
  * (missing `content` array → crash on reload). We coerce at the single boundary where untyped
  * results enter the agent loop, so every downstream consumer can rely on the type.
  */
-function coerceToolResult(raw: unknown): { result: AgentToolResult<any>; malformed: boolean } {
+function coerceToolResult(raw: unknown): { result: AgentToolResult<unknown>; malformed: boolean } {
 	const rawObj = raw && typeof raw === "object" ? (raw as Record<string, unknown>) : null;
 	const rawContent = rawObj?.content;
 	const details = rawObj && "details" in rawObj ? rawObj.details : {};
@@ -98,8 +168,12 @@ function coerceToolResult(raw: unknown): { result: AgentToolResult<any>; malform
 	}
 
 	const content: AgentToolResult["content"] = [];
+	let invalidBlocks = 0;
 	for (const block of rawContent) {
-		if (!block || typeof block !== "object" || !("type" in block)) continue;
+		if (!block || typeof block !== "object" || !("type" in block)) {
+			invalidBlocks++;
+			continue;
+		}
 		if (block.type === "text" && typeof (block as { text?: unknown }).text === "string") {
 			content.push({ type: "text", text: sanitizeText((block as { text: string }).text) });
 		} else if (
@@ -108,9 +182,20 @@ function coerceToolResult(raw: unknown): { result: AgentToolResult<any>; malform
 			typeof (block as { mimeType?: unknown }).mimeType === "string"
 		) {
 			content.push(block as { type: "image"; data: string; mimeType: string });
+		} else {
+			invalidBlocks++;
 		}
 	}
-	return { result: { content, details, ...(explicitError ? { isError: true } : {}) }, malformed: false };
+	if (invalidBlocks > 0) {
+		content.push({
+			type: "text",
+			text: `Tool returned an invalid result: ${invalidBlocks} content block${invalidBlocks === 1 ? "" : "s"} had an unsupported shape.`,
+		});
+	}
+	return {
+		result: { content, details, ...(explicitError || invalidBlocks > 0 ? { isError: true } : {}) },
+		malformed: invalidBlocks > 0,
+	};
 }
 
 /**
@@ -176,7 +261,7 @@ export function agentLoopContinue(
 
 	(async () => {
 		const newMessages: AgentMessage[] = [];
-		const currentContext: AgentContext = { ...context };
+		const currentContext: AgentContext = { ...context, messages: [...context.messages] };
 
 		stream.push({ type: "agent_start" });
 		stream.push({ type: "turn_start" });
@@ -211,9 +296,6 @@ function buildAgentEndEvent(
 ): Extract<AgentEvent, { type: "agent_end" }> {
 	if (!telemetry) return { type: "agent_end", messages };
 	const snapshot = telemetry.collector.snapshot({ stepCount });
-	if (telemetry.collector.markRunEnded()) {
-		fireOnRunEnd(telemetry, snapshot.summary, snapshot.coverage);
-	}
 	return { type: "agent_end", messages, telemetry: snapshot.summary, coverage: snapshot.coverage };
 }
 
@@ -313,22 +395,26 @@ function normalizeMessagesForProvider(
 		return messages;
 	}
 
-	let changed = false;
-	const normalized = messages.map(message => {
+	let hasThinking = false;
+	for (const message of messages) {
+		if (message.role !== "assistant" || !Array.isArray(message.content)) continue;
+		for (const block of message.content) {
+			if (block.type === "thinking") {
+				hasThinking = true;
+				break;
+			}
+		}
+		if (hasThinking) break;
+	}
+	if (!hasThinking) return messages;
+
+	return messages.map(message => {
 		if (message.role !== "assistant" || !Array.isArray(message.content)) {
 			return message;
 		}
-
 		const filtered = message.content.filter(block => block.type !== "thinking");
-		if (filtered.length === message.content.length) {
-			return message;
-		}
-
-		changed = true;
-		return { ...message, content: filtered };
+		return filtered.length === message.content.length ? message : { ...message, content: filtered };
 	});
-
-	return changed ? normalized : messages;
 }
 
 export const INTENT_FIELD = "_i";
@@ -552,6 +638,12 @@ async function runLoopBody(
 					continue;
 				}
 			}
+			if (recovered) {
+				message = snapshotAssistantMessage(message);
+				currentContext.messages.push(message);
+				stream.push({ type: "message_start", message: snapshotAssistantMessage(message) });
+				stream.push({ type: "message_end", message: snapshotAssistantMessage(message) });
+			}
 			newMessages.push(message);
 			let steeringMessagesFromExecution: AgentMessage[] | undefined;
 
@@ -640,6 +732,9 @@ async function runLoopBody(
 						status: "skipped",
 					});
 				}
+				if (message.stopReason === "length" && toolResults.length > 0) {
+					hasMoreToolCalls = true;
+				}
 			}
 
 			stream.push({ type: "turn_end", message, toolResults });
@@ -816,8 +911,27 @@ async function streamAssistantResponse(
 
 			let partialMessage: AssistantMessage | null = null;
 			let addedPartial = false;
+			const completedToolCallIds = new Set<string>();
 
 			const responseIterator = response[Symbol.asyncIterator]();
+			const finishAbortedStream = async (): Promise<AssistantMessage> => {
+				try {
+					await responseIterator.return?.();
+				} catch {
+					// Provider cancellation failures cannot change the committed aborted message.
+				}
+				const aborted = emitAbortedAssistantMessage(
+					partialMessage,
+					addedPartial,
+					completedToolCallIds,
+					context,
+					config,
+					stream,
+					requestSignal,
+				);
+				await finishChat(aborted);
+				return aborted;
+			};
 
 			// Set up a single abort race: register the abort listener once for the whole
 			// stream and reuse the same race promise for every iterator.next() instead of
@@ -826,16 +940,7 @@ async function streamAssistantResponse(
 			let detachAbortListener: (() => void) | undefined;
 			if (requestSignal) {
 				if (requestSignal.aborted) {
-					const aborted = emitAbortedAssistantMessage(
-						partialMessage,
-						addedPartial,
-						context,
-						config,
-						stream,
-						requestSignal,
-					);
-					await finishChat(aborted);
-					return aborted;
+					return await finishAbortedStream();
 				}
 				const { promise, resolve } = Promise.withResolvers<typeof ABORTED>();
 				const onAbort = () => resolve(ABORTED);
@@ -850,37 +955,51 @@ async function streamAssistantResponse(
 					if (abortRacePromise) {
 						const result = await Promise.race([responseIterator.next(), abortRacePromise]);
 						if (result === ABORTED) {
-							responseIterator.return?.()?.catch(() => {});
-							const aborted = emitAbortedAssistantMessage(
-								partialMessage,
-								addedPartial,
-								context,
-								config,
-								stream,
-								requestSignal,
-							);
-							await finishChat(aborted);
-							return aborted;
+							return await finishAbortedStream();
 						}
 						next = result;
 					} else {
 						next = await responseIterator.next();
 					}
-					if (requestSignal?.aborted) {
-						const aborted = emitAbortedAssistantMessage(
-							partialMessage,
-							addedPartial,
-							context,
-							config,
-							stream,
-							requestSignal,
-						);
-						await finishChat(aborted);
-						return aborted;
-					}
 					if (next.done) break;
 
 					const event = next.value;
+					if (event.type === "done" || event.type === "error") {
+						let finalMessage = retainCompletedToolCalls(await response.result(), completedToolCallIds);
+						if (harmonyMitigationEnabled) {
+							const detection = detectHarmonyLeakInAssistantMessage(finalMessage);
+							if (detection) {
+								const recovered = recoverHarmonyToolCall(finalMessage, detection);
+								const removed = recovered?.removed ?? extractHarmonyRemoved(finalMessage, detection);
+								if (addedPartial) {
+									emitDiscardedHarmonyPartial(
+										partialMessage,
+										stream,
+										`Discarded after GPT-5 Harmony protocol leakage (${signalListLabel(detection.signals)})`,
+									);
+									context.messages.pop();
+									addedPartial = false;
+								}
+								throw new HarmonyLeakInterruption(detection, removed, recovered);
+							}
+						}
+						finalMessage = snapshotAssistantMessage(finalMessage);
+						if (addedPartial) {
+							context.messages[context.messages.length - 1] = finalMessage;
+						} else {
+							context.messages.push(finalMessage);
+						}
+						if (!addedPartial) {
+							stream.push({ type: "message_start", message: snapshotAssistantMessage(finalMessage) });
+						}
+						stream.push({ type: "message_end", message: snapshotAssistantMessage(finalMessage) });
+						await finishChat(finalMessage);
+						return finalMessage;
+					}
+					if (requestSignal?.aborted) {
+						return await finishAbortedStream();
+					}
+
 					// Yield to the event loop periodically to prevent busy-wait
 					// when the LLM is streaming chunks faster than the loop can rest.
 					await yieldIfDue();
@@ -890,7 +1009,7 @@ async function streamAssistantResponse(
 							partialMessage = event.partial;
 							context.messages.push(partialMessage);
 							addedPartial = true;
-							stream.push({ type: "message_start", message: { ...partialMessage } });
+							stream.push({ type: "message_start", message: snapshotAssistantMessage(partialMessage) });
 							break;
 
 						case "text_start":
@@ -903,63 +1022,48 @@ async function streamAssistantResponse(
 						case "toolcall_delta":
 						case "toolcall_end":
 							if (partialMessage) {
+								if (event.type === "toolcall_end") {
+									completedToolCallIds.add(event.toolCall.id);
+								}
 								partialMessage = event.partial;
 								context.messages[context.messages.length - 1] = partialMessage;
 								config.onAssistantMessageEvent?.(partialMessage, event);
-								if (signal?.aborted) {
-									continue;
-								}
 								stream.push({
 									type: "message_update",
-									assistantMessageEvent: event,
-									message: { ...partialMessage },
+									assistantMessageEvent: snapshotAssistantMessageEvent(event),
+									message: snapshotAssistantMessage(partialMessage),
 								});
 							}
 							break;
-
-						case "done":
-						case "error": {
-							const finalMessage = await response.result();
-							if (harmonyMitigationEnabled) {
-								const detection = detectHarmonyLeakInAssistantMessage(finalMessage);
-								if (detection) {
-									const removed = extractHarmonyRemoved(finalMessage, detection);
-									if (addedPartial) {
-										context.messages.pop();
-										addedPartial = false;
-									}
-									throw new HarmonyLeakInterruption(detection, removed);
-								}
-							}
-							if (addedPartial) {
-								context.messages[context.messages.length - 1] = finalMessage;
-							} else {
-								context.messages.push(finalMessage);
-							}
-							if (!addedPartial) {
-								stream.push({ type: "message_start", message: { ...finalMessage } });
-							}
-							stream.push({ type: "message_end", message: finalMessage });
-							await finishChat(finalMessage);
-							return finalMessage;
-						}
 					}
 				}
 			} finally {
 				detachAbortListener?.();
 			}
 
-			const trailing = await response.result();
+			let trailing = await response.result();
 			if (harmonyMitigationEnabled) {
 				const detection = detectHarmonyLeakInAssistantMessage(trailing);
 				if (detection) {
+					const recovered = recoverHarmonyToolCall(trailing, detection);
+					const removed = recovered?.removed ?? extractHarmonyRemoved(trailing, detection);
 					if (addedPartial) {
+						emitDiscardedHarmonyPartial(
+							partialMessage,
+							stream,
+							`Discarded after GPT-5 Harmony protocol leakage (${signalListLabel(detection.signals)})`,
+						);
 						context.messages.pop();
 						addedPartial = false;
 					}
-					throw new HarmonyLeakInterruption(detection, extractHarmonyRemoved(trailing, detection));
+					throw new HarmonyLeakInterruption(detection, removed, recovered);
 				}
 			}
+			trailing = snapshotAssistantMessage(trailing);
+			if (addedPartial) {
+				context.messages[context.messages.length - 1] = trailing;
+				stream.push({ type: "message_end", message: snapshotAssistantMessage(trailing) });
+			}
 			await finishChat(trailing);
 			return trailing;
 		});
@@ -973,6 +1077,30 @@ async function streamAssistantResponse(
 	}
 }
 
+function retainCompletedToolCalls(message: AssistantMessage, completedToolCallIds: ReadonlySet<string>): AssistantMessage {
+	if (message.stopReason !== "error" && message.stopReason !== "aborted") return message;
+	let changed = false;
+	const content = message.content.filter(block => {
+		if (block.type !== "toolCall") return true;
+		const keep = completedToolCallIds.has(block.id);
+		if (!keep) changed = true;
+		return keep;
+	});
+	return changed ? { ...message, content } : message;
+}
+
+function emitDiscardedHarmonyPartial(
+	partialMessage: AssistantMessage | null,
+	stream: EventStream<AgentEvent, AgentMessage[]>,
+	errorMessage: string,
+): void {
+	if (!partialMessage) return;
+	stream.push({
+		type: "message_end",
+		message: snapshotAssistantMessage({ ...partialMessage, stopReason: "error", errorMessage }),
+	});
+}
+
 /** Resolve the human-readable reason an abort carried. A caller that aborts via
  *  `AbortController.abort(reason)` with a string or a non-`AbortError` `Error`
  *  (e.g. the coding agent's user-interrupt label) gets that text surfaced on the
@@ -991,39 +1119,45 @@ export function abortReasonText(signal: AbortSignal | undefined): string {
 function emitAbortedAssistantMessage(
 	partialMessage: AssistantMessage | null,
 	addedPartial: boolean,
+	completedToolCallIds: ReadonlySet<string>,
 	context: AgentContext,
 	config: AgentLoopConfig,
 	stream: EventStream<AgentEvent, AgentMessage[]>,
 	requestSignal: AbortSignal | undefined,
 ): AssistantMessage {
 	const errorMessage = abortReasonText(requestSignal);
-	const abortedMessage: AssistantMessage = partialMessage
-		? { ...partialMessage, stopReason: "aborted", errorMessage }
-		: {
-				role: "assistant",
-				content: [],
-				api: config.model.api,
-				provider: config.model.provider,
-				model: config.model.id,
-				usage: {
-					input: 0,
-					output: 0,
-					cacheRead: 0,
-					cacheWrite: 0,
-					totalTokens: 0,
-					cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, total: 0 },
-				},
-				stopReason: "aborted",
-				errorMessage,
-				timestamp: Date.now(),
-			};
+	const abortedMessage = snapshotAssistantMessage(
+		retainCompletedToolCalls(
+			partialMessage
+				? { ...partialMessage, stopReason: "aborted", errorMessage }
+				: {
+						role: "assistant",
+						content: [],
+						api: config.model.api,
+						provider: config.model.provider,
+						model: config.model.id,
+						usage: {
+							input: 0,
+							output: 0,
+							cacheRead: 0,
+							cacheWrite: 0,
+							totalTokens: 0,
+							cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, total: 0 },
+						},
+						stopReason: "aborted",
+						errorMessage,
+						timestamp: Date.now(),
+					},
+			completedToolCallIds,
+		),
+	);
 	if (addedPartial) {
 		context.messages[context.messages.length - 1] = abortedMessage;
 	} else {
 		context.messages.push(abortedMessage);
-		stream.push({ type: "message_start", message: { ...abortedMessage } });
+		stream.push({ type: "message_start", message: snapshotAssistantMessage(abortedMessage) });
 	}
-	stream.push({ type: "message_end", message: abortedMessage });
+	stream.push({ type: "message_end", message: snapshotAssistantMessage(abortedMessage) });
 	return abortedMessage;
 }
 
@@ -1061,7 +1195,7 @@ async function executeToolCalls(
 		: steeringAbortController.signal;
 	const interruptState = { triggered: false };
 	let steeringMessages: AgentMessage[] | undefined;
-	let steeringCheck: Promise<void> | null = null;
+	let steeringCheckTail: Promise<void> = Promise.resolve();
 
 	const records = toolCalls.map(toolCall => ({
 		toolCall,
@@ -1085,21 +1219,17 @@ async function executeToolCalls(
 		if (!shouldInterruptImmediately || !getSteeringMessages || interruptState.triggered) {
 			return;
 		}
-		if (steeringCheck) {
-			await steeringCheck;
-			return;
-		}
-		steeringCheck = (async () => {
+		const check = steeringCheckTail.then(async () => {
+			if (interruptState.triggered) return;
 			const steering = await getSteeringMessages();
 			if (steering.length > 0) {
 				steeringMessages = steering;
 				interruptState.triggered = true;
 				steeringAbortController.abort();
 			}
-		})().finally(() => {
-			steeringCheck = null;
 		});
-		await steeringCheck;
+		steeringCheckTail = check.catch(() => {});
+		await check;
 	};
 
 	const emitToolResult = (record: (typeof records)[number], result: AgentToolResult<any>, isError: boolean): void => {
@@ -1171,6 +1301,16 @@ async function executeToolCalls(
 			}
 		}
 		record.args = argsForExecution;
+		if (toolSignal.aborted) {
+			record.skipped = true;
+			recordSkippedTool(telemetry, {
+				toolCallId: toolCall.id,
+				toolName: toolCall.name,
+				status: "aborted",
+			});
+			emitToolResult(record, createToolSignalAbortedResult(toolSignal), true);
+			return;
+		}
 		record.started = true;
 		stream.push({
 			type: "tool_execution_start",
@@ -1198,6 +1338,11 @@ async function executeToolCalls(
 		await runInActiveSpan(toolSpan, async () => {
 			try {
 				if (!tool) throw new Error(`Tool ${toolCall.name} not found`);
+				if (toolSignal.aborted) {
+					result = createToolSignalAbortedResult(toolSignal);
+					isError = true;
+					return;
+				}
 
 				let effectiveArgs: Record<string, unknown>;
 				try {
@@ -1224,8 +1369,15 @@ async function executeToolCalls(
 						throw new ToolCallBlockedError(beforeResult.reason);
 					}
 				}
-				// Reflect post-hook args so emitted tool results / afterToolCall see what actually executed.
-				record.args = effectiveArgs;
+				if (toolSignal.aborted) {
+					result = createToolSignalAbortedResult(toolSignal);
+					isError = true;
+					return;
+				}
+				const executionArgs = transformToolCallArguments
+					? transformToolCallArguments(effectiveArgs, toolCall.name)
+					: effectiveArgs;
+				record.args = executionArgs;
 
 				const toolContext = getToolContext
 					? getToolContext({
@@ -1237,14 +1389,14 @@ async function executeToolCalls(
 					: undefined;
 				const rawResult = await tool.execute(
 					toolCall.id,
-					transformToolCallArguments ? transformToolCallArguments(effectiveArgs, toolCall.name) : effectiveArgs,
+					executionArgs,
 					toolSignal,
 					partialResult => {
 						stream.push({
 							type: "tool_execution_update",
 							toolCallId: toolCall.id,
 							toolName: toolCall.name,
-							args: effectiveArgs,
+							args: executionArgs,
 							partialResult: coerceToolResult(partialResult).result,
 						});
 					},
@@ -1262,7 +1414,7 @@ async function executeToolCalls(
 				isError = true;
 			}
 
-			if (afterToolCall) {
+			if (afterToolCall && !toolSignal.aborted) {
 				try {
 					const after = await afterToolCall(
 						{
@@ -1295,6 +1447,7 @@ async function executeToolCalls(
 		});
 
 		const interrupted = interruptState.triggered;
+		const abortedDuringExecution = toolSignal.aborted && isError;
 		if (interrupted) {
 			record.skipped = true;
 			emitToolResult(record, createSkippedToolResult(), true);
@@ -1305,13 +1458,14 @@ async function executeToolCalls(
 		const firstTextBlock = result.content?.[0];
 		const errorMessageForSpan =
 			caughtError === undefined && isError && firstTextBlock?.type === "text" ? firstTextBlock.text : undefined;
-		const status = interrupted
-			? "aborted"
-			: caughtError instanceof ToolCallBlockedError
-				? "blocked"
-				: isError
-					? "error"
-					: "ok";
+		const status =
+			interrupted || abortedDuringExecution
+				? "aborted"
+				: caughtError instanceof ToolCallBlockedError
+					? "blocked"
+					: isError
+						? "error"
+						: "ok";
 		finishExecuteToolSpan(telemetry, toolSpan, {
 			result,
 			isError,
@@ -1417,6 +1571,14 @@ function createAbortedToolResult(
 	return toolResultMessage;
 }
 
+function createToolSignalAbortedResult(signal: AbortSignal): AgentToolResult<unknown> {
+	const reason = abortReasonText(signal);
+	return {
+		content: [{ type: "text", text: `Tool was not executed because the run was aborted: ${reason}.` }],
+		details: {},
+	};
+}
+
 function createSkippedToolResult(): AgentToolResult<any> {
 	return {
 		content: [{ type: "text", text: "Skipped due to queued user message." }],
diff --git a/packages/ai/CHANGELOG.md b/packages/ai/CHANGELOG.md
index b67b895bc..272a04f91 100644
--- a/packages/ai/CHANGELOG.md
+++ b/packages/ai/CHANGELOG.md
@@ -1,13 +1,17 @@
 # Changelog
 
 ## [Unreleased]
+### Changed
+
+- Changed Anthropic retry handling to avoid retrying 4xx responses other than 408 and 429
 
 ### Fixed
 
+- Fixed raw Anthropic SSE handling by parsing event frames with strict JSON parsing and matching event-type validation, surfacing malformed frames as stream errors instead of repairing them
+- Fixed Anthropic stream envelope handling to reject duplicate `content_block_start` indexes and block deltas/stops for unopened blocks, preventing malformed envelope states from producing partial output
+- Fixed Anthropic image conversion to normalize `image/jpg` to `image/jpeg` and emit a placeholder for unsupported image MIME types
+- Fixed Anthropic thinking request preparation by clamping `max_tokens` to provider/model limits and adjusting thinking budgets to a valid value
 - Fixed the Anthropic stream parser shipping a truncated tool call as a completed turn. When a transport drop cut the SSE stream mid-`tool_use` and a transparent reconnect spliced a fresh message envelope onto the same stream, the duplicate `message_start` was deduped but the orphaned tool block — which never received its `content_block_stop` — survived in the assistant message with its seed `{}` (or partially-parsed) arguments. The terminal stop signal from the reconnect then let it flow through as a normal tool call, so e.g. a `read` dispatched with `{}` failed downstream validation (`path: expected string, received undefined`). The parser now treats any tool block left open at stream end as a truncated envelope and routes it through the existing retry/error path instead of emitting bogus arguments.
-
-### Fixed
-
 - Fixed the Zhipu Coding Plan login prompt advertising a misleading `sk-...` placeholder. Zhipu API keys are formatted `<id>.<secret>` (no `sk-` prefix), so the placeholder now matches the actual format instead of suggesting the wrong shape. ([#2106](https://github.com/can1357/oh-my-pi/issues/2106))
 
 ## [15.10.4] - 2026-06-08
diff --git a/packages/ai/src/providers/anthropic.ts b/packages/ai/src/providers/anthropic.ts
index 20a5fad31..652a854cd 100644
--- a/packages/ai/src/providers/anthropic.ts
+++ b/packages/ai/src/providers/anthropic.ts
@@ -57,7 +57,7 @@ import { AssistantMessageEventStream } from "../utils/event-stream";
 import { isFoundryEnabled } from "../utils/foundry";
 import { finalizeErrorMessage, type RawHttpRequestDump, rewriteCopilotError } from "../utils/http-inspector";
 import { getStreamFirstEventTimeoutMs, getStreamIdleTimeoutMs, iterateWithIdleTimeout } from "../utils/idle-iterator";
-import { parseJsonWithRepair, parseStreamingJson, parseStreamingJsonThrottled } from "../utils/json-parse";
+import { parseStreamingJsonThrottled } from "../utils/json-parse";
 import { parseGitHubCopilotApiKey } from "../utils/oauth/github-copilot";
 import { notifyProviderResponse } from "../utils/provider-response";
 import { isCopilotTransientModelError } from "../utils/retry";
@@ -257,6 +257,26 @@ export function buildAnthropicHeaders(options: AnthropicHeaderOptions): Record<s
 }
 
 type AnthropicCacheControl = NonNullable<TextBlockParam["cache_control"]>;
+type AnthropicImageMediaType = "image/jpeg" | "image/png" | "image/gif" | "image/webp";
+
+function normalizeAnthropicImageMediaType(mimeType: string): AnthropicImageMediaType | undefined {
+	const normalized = mimeType.trim().toLowerCase();
+	if (normalized === "image/jpg") return "image/jpeg";
+	if (
+		normalized === "image/jpeg" ||
+		normalized === "image/png" ||
+		normalized === "image/gif" ||
+		normalized === "image/webp"
+	) {
+		return normalized;
+	}
+	return undefined;
+}
+
+function cloneAnthropicCacheControl(cacheControl: AnthropicCacheControl): AnthropicCacheControl {
+	return { ...cacheControl };
+}
+
 
 type AnthropicOutputConfig = NonNullable<MessageCreateParamsStreaming["output_config"]>;
 
@@ -750,42 +770,67 @@ function convertContentBlocks(
 					type: "image";
 					source: {
 						type: "base64";
-						media_type: "image/jpeg" | "image/png" | "image/gif" | "image/webp";
+						media_type: AnthropicImageMediaType;
 						data: string;
 					};
 			  }
 	  > {
-	const textBlocks = content
-		.filter((block): block is TextContent => block.type === "text")
-		.map(block => block.text.toWellFormed())
-		.filter(text => text.trim().length > 0);
-	const imageBlocks = content.filter((block): block is ImageContent => block.type === "image");
-	const omittedImages = !supportsImages && imageBlocks.length > 0;
-	if (imageBlocks.length === 0 || !supportsImages) {
-		if (omittedImages) {
-			textBlocks.push(NON_VISION_IMAGE_PLACEHOLDER);
-		}
-		return textBlocks.join("\n").toWellFormed();
-	}
+	const blocks: Array<
+		| { type: "text"; text: string }
+		| {
+				type: "image";
+				source: {
+					type: "base64";
+					media_type: AnthropicImageMediaType;
+					data: string;
+				};
+		  }
+	> = [];
+	let sawText = false;
+	let sawImage = false;
 
-	const blocks = [
-		...textBlocks.map(text => ({
-			type: "text" as const,
-			text,
-		})),
-		...imageBlocks.map(block => ({
-			type: "image" as const,
+	for (const block of content) {
+		if (block.type === "text") {
+			const text = block.text.toWellFormed();
+			if (text.trim().length === 0) continue;
+			sawText = true;
+			blocks.push({ type: "text", text });
+			continue;
+		}
+
+		if (!supportsImages) {
+			blocks.push({ type: "text", text: NON_VISION_IMAGE_PLACEHOLDER });
+			continue;
+		}
+
+		const mediaType = normalizeAnthropicImageMediaType(block.mimeType);
+		if (!mediaType) {
+			blocks.push({ type: "text", text: `[unsupported image: ${block.mimeType}]` });
+			continue;
+		}
+
+		sawImage = true;
+		blocks.push({
+			type: "image",
 			source: {
-				type: "base64" as const,
-				media_type: block.mimeType as "image/jpeg" | "image/png" | "image/gif" | "image/webp",
+				type: "base64",
+				media_type: mediaType,
 				data: block.data,
 			},
-		})),
-	];
+		});
+	}
 
-	if (!textBlocks.length) {
+	if (!supportsImages) {
+		return blocks
+			.filter((block): block is { type: "text"; text: string } => block.type === "text")
+			.map(block => block.text)
+			.join("\n")
+			.toWellFormed();
+	}
+
+	if (sawImage && !sawText) {
 		blocks.unshift({
-			type: "text" as const,
+			type: "text",
 			text: "(see attached image)",
 		});
 	}
@@ -887,6 +932,16 @@ type FoundryTlsOptions = {
 	key?: string;
 };
 
+const foundryTlsOptionsCache = new Map<string, FoundryTlsOptions | undefined>();
+
+function foundryTlsOptionsCacheKey(): string {
+	return JSON.stringify([
+		$env.NODE_EXTRA_CA_CERTS ?? null,
+		$env.CLAUDE_CODE_CLIENT_CERT ?? null,
+		$env.CLAUDE_CODE_CLIENT_KEY ?? null,
+	]);
+}
+
 function resolveAnthropicBaseUrl(model: Model<"anthropic-messages">, apiKey?: string): string | undefined {
 	if (model.provider === "github-copilot") {
 		return normalizeAnthropicBaseUrl(resolveGitHubCopilotBaseUrl(model.baseUrl, apiKey) ?? model.baseUrl);
@@ -975,6 +1030,9 @@ function resolveFoundryTlsOptions(model: Model<"anthropic-messages">): FoundryTl
 	if (model.provider !== "anthropic") return undefined;
 	if (!isFoundryEnabled()) return undefined;
 
+	const cacheKey = foundryTlsOptionsCacheKey();
+	if (foundryTlsOptionsCache.has(cacheKey)) return foundryTlsOptionsCache.get(cacheKey);
+
 	const ca = resolvePemValue($env.NODE_EXTRA_CA_CERTS, "NODE_EXTRA_CA_CERTS");
 	const cert = resolvePemValue($env.CLAUDE_CODE_CLIENT_CERT, "CLAUDE_CODE_CLIENT_CERT");
 	const key = resolvePemValue($env.CLAUDE_CODE_CLIENT_KEY, "CLAUDE_CODE_CLIENT_KEY");
@@ -987,7 +1045,9 @@ function resolveFoundryTlsOptions(model: Model<"anthropic-messages">): FoundryTl
 	if (ca) options.ca = [...tls.rootCertificates, ca];
 	if (cert) options.cert = cert;
 	if (key) options.key = key;
-	return Object.keys(options).length > 0 ? options : undefined;
+	const resolved = Object.keys(options).length > 0 ? options : undefined;
+	foundryTlsOptionsCache.set(cacheKey, resolved);
+	return resolved;
 }
 
 function buildClaudeCodeTlsFetchOptions(
@@ -1037,14 +1097,8 @@ const ANTHROPIC_MESSAGE_EVENTS: ReadonlySet<string> = new Set([
 ]);
 
 /**
- * Anthropic keepalive `ping` events carry no message content, but they prove the
- * upstream connection is alive during long server-side gaps (extended thinking,
- * slow tool execution). They are normally dropped before reaching the consumer;
- * we instead surface them as lightweight markers so the idle watchdog
- * (`iterateWithIdleTimeout`) resets its deadline on every ping. Without this, a
- * connection that is demonstrably still streaming pings still trips
- * "Anthropic stream stalled while waiting for the next event". The message-event
- * branches in `streamAnthropic` match none of these markers, so they are ignored.
+ * Iterate over Anthropic SSE events from a raw Response, preserving ping events
+ * for liveness and rejecting malformed complete event envelopes.
  */
 type RawMessagePingEvent = { type: "ping" };
 type AnthropicStreamEvent = RawMessageStreamEvent | RawMessagePingEvent;
@@ -1079,7 +1133,10 @@ async function* iterateAnthropicEvents(
 		}
 
 		try {
-			const event = parseJsonWithRepair<RawMessageStreamEvent>(sse.data);
+			const event = JSON.parse(sse.data) as RawMessageStreamEvent;
+			if (event.type !== sse.event) {
+				throw new Error(`event type ${event.type} does not match SSE event ${sse.event}`);
+			}
 			if (event.type === "message_start") {
 				sawMessageStart = true;
 			} else if (event.type === "message_stop") {
@@ -1094,7 +1151,7 @@ async function* iterateAnthropicEvents(
 		}
 	}
 
-	if (sawMessageStart && !sawMessageEnd) {
+	if (sawMessageStart && !sawMessageEnd && !signal?.aborted) {
 		throw createAnthropicStreamEnvelopeError("stream ended before message_stop");
 	}
 }
@@ -1177,17 +1234,12 @@ function getAnthropicCompat(
 const PROVIDER_MAX_RETRIES = 3;
 const PROVIDER_BASE_DELAY_MS = 2000;
 
-/**
- * Check if an error from the Anthropic SDK is a rate-limit/transient error that
- * should be retried before any content has been emitted.
- *
- * Includes malformed JSON stream-envelope parse errors seen from some
- * Anthropic-compatible proxy endpoints.
- */
 /** Transient stream corruption errors where the response was truncated mid-JSON. */
 function isTransientStreamParseError(error: unknown): boolean {
 	if (!(error instanceof Error)) return false;
-	return /json parse error|unterminated string|unexpected end of json input/i.test(error.message);
+	return /unterminated string|unexpected end of json input|unexpected end of data|unexpected eof|end of file|eof while parsing|truncated/i.test(
+		error.message,
+	);
 }
 
 const ANTHROPIC_STREAM_ENVELOPE_ERROR_PREFIX = "Anthropic stream envelope error:";
@@ -1235,6 +1287,8 @@ export function isProviderRetryableError(error: unknown, provider?: string): boo
 	// `streamSimple` a/b/c policy), so surface them immediately instead of
 	// burning the retry budget here.
 	if (isUsageLimitError(error.message)) return false;
+	const status = extractHttpStatusFromError(error);
+	if (status !== undefined && status >= 400 && status < 500 && status !== 408 && status !== 429) return false;
 	const msg = error.message.toLowerCase();
 	if (
 		isUnexpectedSocketCloseMessage(msg) ||
@@ -1268,13 +1322,12 @@ export type AnthropicUsageLike = {
 
 /**
  * Capture Anthropic's optional cache-creation TTL breakdown and server-tool-use
- * counters into the harness Usage shape. Only sets fields that were reported, so
- * a `message_delta` that omits `cache_creation` does not clobber the breakdown
- * established at `message_start`.
+ * counters into the harness Usage shape. Omitted/null fields are no-ops; explicit
+ * zero-valued objects clear prior extras from earlier stream usage snapshots.
  */
 export function applyAnthropicUsageExtras(usage: Usage, source: AnthropicUsageLike): void {
 	const cacheCreation = source.cache_creation;
-	if (cacheCreation) {
+	if (cacheCreation != null) {
 		const fiveMinute = cacheCreation.ephemeral_5m_input_tokens ?? 0;
 		const oneHour = cacheCreation.ephemeral_1h_input_tokens ?? 0;
 		if (fiveMinute > 0 || oneHour > 0) {
@@ -1282,10 +1335,12 @@ export function applyAnthropicUsageExtras(usage: Usage, source: AnthropicUsageLi
 				...(fiveMinute > 0 ? { ephemeral5m: fiveMinute } : {}),
 				...(oneHour > 0 ? { ephemeral1h: oneHour } : {}),
 			};
+		} else {
+			delete usage.cttl;
 		}
 	}
 	const serverToolUse = source.server_tool_use;
-	if (serverToolUse) {
+	if (serverToolUse != null) {
 		const webSearch = serverToolUse.web_search_requests ?? 0;
 		const webFetch = serverToolUse.web_fetch_requests ?? 0;
 		if (webSearch > 0 || webFetch > 0) {
@@ -1293,6 +1348,8 @@ export function applyAnthropicUsageExtras(usage: Usage, source: AnthropicUsageLi
 				...(webSearch > 0 ? { webSearch } : {}),
 				...(webFetch > 0 ? { webFetch } : {}),
 			};
+		} else {
+			delete usage.server;
 		}
 	}
 }
@@ -1466,6 +1523,11 @@ export const streamAnthropic: StreamFunction<"anthropic-messages"> = (
 					let sawEvent = false;
 					let sawMessageStart = false;
 					let sawTerminalEnvelope = false;
+					let sawMessageStop = false;
+					const openBlocks = new Map<
+						number,
+						{ contentIndex: number; kind: "text" | "thinking" | "redactedThinking" | "toolCall" }
+					>();
 
 					const timedAnthropicStream = iterateWithIdleTimeout(anthropicStream, {
 						idleTimeoutMs,
@@ -1508,6 +1570,12 @@ export const streamAnthropic: StreamFunction<"anthropic-messages"> = (
 						}
 
 						if (event.type === "content_block_start") {
+							if (sawTerminalEnvelope) {
+								throw createAnthropicStreamEnvelopeError(`received ${event.type} after terminal stop signal`);
+							}
+							if (openBlocks.has(event.index)) {
+								throw createAnthropicStreamEnvelopeError(`duplicate content_block_start index ${event.index}`);
+							}
 							if (!firstTokenTime) firstTokenTime = Date.now();
 							if (event.content_block.type === "text") {
 								streamedReplayUnsafeContent = true;
@@ -1517,12 +1585,15 @@ export const streamAnthropic: StreamFunction<"anthropic-messages"> = (
 									index: event.index,
 								};
 								output.content.push(block);
+								const contentIndex = output.content.length - 1;
+								openBlocks.set(event.index, { contentIndex, kind: "text" });
 								stream.push({
 									type: "text_start",
-									contentIndex: output.content.length - 1,
+									contentIndex,
 									partial: output,
 								});
 							} else if (event.content_block.type === "thinking") {
+								streamedReplayUnsafeContent = true;
 								const block: Block = {
 									type: "thinking",
 									thinking: "",
@@ -1530,18 +1601,25 @@ export const streamAnthropic: StreamFunction<"anthropic-messages"> = (
 									index: event.index,
 								};
 								output.content.push(block);
+								const contentIndex = output.content.length - 1;
+								openBlocks.set(event.index, { contentIndex, kind: "thinking" });
 								stream.push({
 									type: "thinking_start",
-									contentIndex: output.content.length - 1,
+									contentIndex,
 									partial: output,
 								});
 							} else if (event.content_block.type === "redacted_thinking") {
+								streamedReplayUnsafeContent = true;
 								const block: Block = {
 									type: "redactedThinking",
 									data: event.content_block.data,
 									index: event.index,
 								};
 								output.content.push(block);
+								openBlocks.set(event.index, {
+									contentIndex: output.content.length - 1,
+									kind: "redactedThinking",
+								});
 							} else if (event.content_block.type === "tool_use") {
 								streamedReplayUnsafeContent = true;
 								const block: Block = {
@@ -1555,92 +1633,115 @@ export const streamAnthropic: StreamFunction<"anthropic-messages"> = (
 									index: event.index,
 								};
 								output.content.push(block);
+								const contentIndex = output.content.length - 1;
+								openBlocks.set(event.index, { contentIndex, kind: "toolCall" });
 								stream.push({
 									type: "toolcall_start",
-									contentIndex: output.content.length - 1,
+									contentIndex,
 									partial: output,
 								});
 							}
 						} else if (event.type === "content_block_delta") {
+							if (sawTerminalEnvelope) {
+								throw createAnthropicStreamEnvelopeError(`received ${event.type} after terminal stop signal`);
+							}
+							const openBlock = openBlocks.get(event.index);
+							if (!openBlock) {
+								throw createAnthropicStreamEnvelopeError(`received content_block_delta for unopened index ${event.index}`);
+							}
+							const block = blocks[openBlock.contentIndex];
 							if (event.delta.type === "text_delta") {
-								const index = blocks.findIndex(b => b.index === event.index);
-								const block = blocks[index];
-								if (block && block.type === "text") {
-									block.text += event.delta.text;
-									stream.push({
-										type: "text_delta",
-										contentIndex: index,
-										delta: event.delta.text,
-										partial: output,
-									});
+								if (openBlock.kind !== "text" || block?.type !== "text") {
+									throw createAnthropicStreamEnvelopeError(`received text_delta for ${openBlock.kind} block`);
 								}
+								streamedReplayUnsafeContent = true;
+								block.text += event.delta.text;
+								stream.push({
+									type: "text_delta",
+									contentIndex: openBlock.contentIndex,
+									delta: event.delta.text,
+									partial: output,
+								});
 							} else if (event.delta.type === "thinking_delta") {
-								const index = blocks.findIndex(b => b.index === event.index);
-								const block = blocks[index];
-								if (block && block.type === "thinking") {
-									block.thinking += event.delta.thinking;
-									stream.push({
-										type: "thinking_delta",
-										contentIndex: index,
-										delta: event.delta.thinking,
-										partial: output,
-									});
+								if (openBlock.kind !== "thinking" || block?.type !== "thinking") {
+									throw createAnthropicStreamEnvelopeError(`received thinking_delta for ${openBlock.kind} block`);
 								}
+								streamedReplayUnsafeContent = true;
+								block.thinking += event.delta.thinking;
+								stream.push({
+									type: "thinking_delta",
+									contentIndex: openBlock.contentIndex,
+									delta: event.delta.thinking,
+									partial: output,
+								});
 							} else if (event.delta.type === "input_json_delta") {
-								const index = blocks.findIndex(b => b.index === event.index);
-								const block = blocks[index];
-								if (block && block.type === "toolCall") {
-									block.partialJson += event.delta.partial_json;
-									const throttled = parseStreamingJsonThrottled(block.partialJson, block.lastParseLen ?? 0);
-									if (throttled) {
-										block.arguments = throttled.value;
-										block.lastParseLen = throttled.parsedLen;
-									}
-									stream.push({
-										type: "toolcall_delta",
-										contentIndex: index,
-										delta: event.delta.partial_json,
-										partial: output,
-									});
+								if (openBlock.kind !== "toolCall" || block?.type !== "toolCall") {
+									throw createAnthropicStreamEnvelopeError(`received input_json_delta for ${openBlock.kind} block`);
 								}
+								streamedReplayUnsafeContent = true;
+								block.partialJson += event.delta.partial_json;
+								const throttled = parseStreamingJsonThrottled(block.partialJson, block.lastParseLen ?? 0);
+								if (throttled) {
+									block.arguments = throttled.value;
+									block.lastParseLen = throttled.parsedLen;
+								}
+								stream.push({
+									type: "toolcall_delta",
+									contentIndex: openBlock.contentIndex,
+									delta: event.delta.partial_json,
+									partial: output,
+								});
 							} else if (event.delta.type === "signature_delta") {
-								const index = blocks.findIndex(b => b.index === event.index);
-								const block = blocks[index];
-								if (block && block.type === "thinking") {
-									block.thinkingSignature = block.thinkingSignature || "";
-									block.thinkingSignature += event.delta.signature;
+								if (openBlock.kind !== "thinking" || block?.type !== "thinking") {
+									throw createAnthropicStreamEnvelopeError(`received signature_delta for ${openBlock.kind} block`);
 								}
+								streamedReplayUnsafeContent = true;
+								block.thinkingSignature = block.thinkingSignature || "";
+								block.thinkingSignature += event.delta.signature;
 							}
 						} else if (event.type === "content_block_stop") {
-							const index = blocks.findIndex(b => b.index === event.index);
-							const block = blocks[index];
-							if (block) {
-								delete (block as { index?: number }).index;
-								if (block.type === "text") {
-									stream.push({
-										type: "text_end",
-										contentIndex: index,
-										content: block.text,
-										partial: output,
-									});
-								} else if (block.type === "thinking") {
-									stream.push({
-										type: "thinking_end",
-										contentIndex: index,
-										content: block.thinking,
-										partial: output,
-									});
-								} else if (block.type === "toolCall") {
-									block.arguments = parseStreamingJson(block.partialJson);
-									delete (block as { partialJson?: string }).partialJson;
-									delete (block as { lastParseLen?: number }).lastParseLen;
-									stream.push({
-										type: "toolcall_end",
-										contentIndex: index,
-										toolCall: block,
-										partial: output,
-									});
-								}
+							if (sawTerminalEnvelope) {
+								throw createAnthropicStreamEnvelopeError(`received ${event.type} after terminal stop signal`);
+							}
+							const openBlock = openBlocks.get(event.index);
+							if (!openBlock) {
+								throw createAnthropicStreamEnvelopeError(`received content_block_stop for unopened index ${event.index}`);
+							}
+							const block = blocks[openBlock.contentIndex];
+							if (!block || block.type !== openBlock.kind) {
+								throw createAnthropicStreamEnvelopeError(`content_block_stop kind mismatch for index ${event.index}`);
+							}
+							openBlocks.delete(event.index);
+							delete (block as { index?: number }).index;
+							if (block.type === "text") {
+								streamedReplayUnsafeContent = true;
+								stream.push({
+									type: "text_end",
+									contentIndex: openBlock.contentIndex,
+									content: block.text,
+									partial: output,
+								});
+							} else if (block.type === "thinking") {
+								streamedReplayUnsafeContent = true;
+								stream.push({
+									type: "thinking_end",
+									contentIndex: openBlock.contentIndex,
+									content: block.thinking,
+									partial: output,
+								});
+							} else if (block.type === "toolCall") {
+								streamedReplayUnsafeContent = true;
+								const finalJson =
+									block.partialJson.length > 0 ? block.partialJson : JSON.stringify(block.arguments ?? {});
+								block.arguments = JSON.parse(finalJson) as ToolCall["arguments"];
+								delete (block as { partialJson?: string }).partialJson;
+								delete (block as { lastParseLen?: number }).lastParseLen;
+								stream.push({
+									type: "toolcall_end",
+									contentIndex: openBlock.contentIndex,
+									toolCall: block,
+									partial: output,
+								});
 							}
 						} else if (event.type === "message_delta") {
 							const rawStopReason = event.delta.stop_reason;
@@ -1683,6 +1784,7 @@ export const streamAnthropic: StreamFunction<"anthropic-messages"> = (
 							calculateCost(model, output.usage);
 						} else if (event.type === "message_stop") {
 							sawTerminalEnvelope = true;
+							sawMessageStop = true;
 						}
 					}
 
@@ -1696,25 +1798,18 @@ export const streamAnthropic: StreamFunction<"anthropic-messages"> = (
 					if (!sawEvent || !sawMessageStart) {
 						throw createAnthropicStreamEnvelopeError("stream ended before message_start");
 					}
-					if (!sawTerminalEnvelope) {
-						throw createAnthropicStreamEnvelopeError("stream ended before terminal stop signal");
+					if (!sawMessageStop) {
+						throw createAnthropicStreamEnvelopeError("stream ended before message_stop");
 					}
-
-					// An open tool_use block — one that never received its
-					// `content_block_stop` — means the stream was truncated mid-tool-call.
-					// In practice this is a transport drop that a transparent reconnect
-					// splices back together: the reconnect's `message_start` is deduped
-					// above, yet the orphaned block survives with its seed `{}` (or a
-					// partial-parse) arguments. Emitting it would dispatch a tool call the
-					// model never finished generating. Surface it as a truncated envelope so
-					// the existing retry/error path engages instead of shipping bogus args.
-					if (
-						blocks.some(
-							block =>
-								block.type === "toolCall" && (block as { partialJson?: string }).partialJson !== undefined,
-						)
-					) {
-						throw createAnthropicStreamEnvelopeError("stream ended with an unterminated tool_use block");
+					if (openBlocks.size > 0) {
+						const firstOpenBlock = openBlocks.entries().next().value;
+						if (firstOpenBlock) {
+							const [openIndex, openBlock] = firstOpenBlock;
+							throw createAnthropicStreamEnvelopeError(
+								`stream ended with an unterminated ${openBlock.kind} block at index ${openIndex}`,
+							);
+						}
+						throw createAnthropicStreamEnvelopeError("stream ended with an unterminated content block");
 					}
 
 					if (output.stopReason === "aborted" || output.stopReason === "error") {
@@ -1849,12 +1944,11 @@ function applyClaudeCodeSystemCache(
 	blocks: AnthropicSystemBlock[],
 	cacheControl: AnthropicCacheControl | undefined,
 ): number {
-	if (!cacheControl || blocks.length <= 2) return 0;
-	blocks[2] = { ...blocks[2], cache_control: cacheControl };
-	if (blocks.length === 3) return 1;
+	if (!cacheControl || blocks.length === 0) return 0;
 	const lastIndex = blocks.length - 1;
-	blocks[lastIndex] = { ...blocks[lastIndex], cache_control: cacheControl };
-	return 2;
+	if (blocks[lastIndex].cache_control != null) return 0;
+	blocks[lastIndex] = { ...blocks[lastIndex], cache_control: cloneAnthropicCacheControl(cacheControl) };
+	return 1;
 }
 
 export function buildAnthropicSystemBlocks(
@@ -1891,8 +1985,8 @@ export function buildAnthropicSystemBlocks(
 		blocks.push({ type: "text", text: prompt });
 	}
 	const lastIndex = blocks.length - 1;
-	if (cacheControl && lastIndex >= 0) {
-		blocks[lastIndex] = { ...blocks[lastIndex], cache_control: cacheControl };
+	if (cacheControl && lastIndex >= 0 && blocks[lastIndex].cache_control == null) {
+		blocks[lastIndex] = { ...blocks[lastIndex], cache_control: cloneAnthropicCacheControl(cacheControl) };
 	}
 	return blocks.length > 0 ? blocks : undefined;
 }
@@ -2009,7 +2103,6 @@ export function buildAnthropicClientOptions(args: AnthropicClientOptionsArgs): A
 			...(tlsFetchOptions ? { fetchOptions: tlsFetchOptions } : {}),
 		};
 	}
-
 	// OpenCode Zen's Anthropic-compatible gateway accepts bearer auth only;
 	// leaving apiKey set lets the client add X-Api-Key, which upstream Alibaba rejects.
 	if (model.provider === "opencode-zen") {
@@ -2025,9 +2118,16 @@ export function buildAnthropicClientOptions(args: AnthropicClientOptionsArgs): A
 		};
 	}
 
+	const authorizationHeader = getHeaderCaseInsensitive(defaultHeaders, "Authorization");
+	const shouldSuppressClientApiKey =
+		!oauthToken &&
+		!isAnthropicApiBaseUrl(baseUrl) &&
+		typeof authorizationHeader === "string" &&
+		/^Bearer\s+/i.test(authorizationHeader);
+
 	return {
 		isOAuthToken: oauthToken,
-		apiKey: oauthToken ? null : apiKey,
+		apiKey: oauthToken || shouldSuppressClientApiKey ? null : apiKey,
 		authToken: oauthToken ? apiKey : undefined,
 		baseURL: baseUrl,
 		maxRetries: 5,
@@ -2052,6 +2152,7 @@ function disableThinkingIfToolChoiceForced(params: MessageCreateParamsStreaming)
 	if (toolChoice.type !== "any" && toolChoice.type !== "tool") return;
 
 	delete params.thinking;
+	delete params.context_management;
 	const outputConfig = params.output_config as AnthropicOutputConfig | undefined;
 	if (!outputConfig) return;
 
@@ -2068,11 +2169,20 @@ function ensureMaxTokensForThinking(params: MessageCreateParamsStreaming, model:
 	const budgetTokens = thinking.budget_tokens ?? 0;
 	if (budgetTokens <= 0) return;
 
-	const maxTokens = params.max_tokens ?? 0;
-	const requiredMaxTokens = budgetTokens + OUTPUT_FALLBACK_BUFFER;
-	if (maxTokens < requiredMaxTokens) {
-		params.max_tokens = Math.min(requiredMaxTokens, model.maxTokens);
+	const maxAllowedTokens = Math.min(CLAUDE_CODE_MAX_OUTPUT_TOKENS, model.maxTokens);
+	const currentMaxTokens = Math.min(params.max_tokens ?? maxAllowedTokens, maxAllowedTokens);
+	const raisedMaxTokens = Math.min(Math.max(currentMaxTokens, budgetTokens + OUTPUT_FALLBACK_BUFFER), maxAllowedTokens);
+	params.max_tokens = raisedMaxTokens;
+
+	if (budgetTokens + OUTPUT_FALLBACK_BUFFER <= raisedMaxTokens) return;
+
+	const clampedBudget = raisedMaxTokens - OUTPUT_FALLBACK_BUFFER;
+	if (clampedBudget <= 0) {
+		throw new Error(
+			`Anthropic thinking budget requires max_tokens greater than ${OUTPUT_FALLBACK_BUFFER}; got ${raisedMaxTokens}`,
+		);
 	}
+	thinking.budget_tokens = clampedBudget;
 }
 
 type CacheControlBlock = {
@@ -2082,39 +2192,35 @@ type CacheControlBlock = {
 function applyCacheControlToLastBlock<T extends CacheControlBlock>(
 	blocks: T[],
 	cacheControl: AnthropicCacheControl,
-): void {
-	if (blocks.length === 0) return;
+): boolean {
+	if (blocks.length === 0) return false;
 	const lastIndex = blocks.length - 1;
-	blocks[lastIndex] = { ...blocks[lastIndex], cache_control: cacheControl };
+	if (blocks[lastIndex].cache_control != null) return false;
+	blocks[lastIndex] = { ...blocks[lastIndex], cache_control: cloneAnthropicCacheControl(cacheControl) };
+	return true;
 }
 
 function applyCacheControlToLastTextBlock(
 	blocks: Array<ContentBlockParam & CacheControlBlock>,
 	cacheControl: AnthropicCacheControl,
-): void {
-	if (blocks.length === 0) return;
+): boolean {
+	if (blocks.length === 0) return false;
 	for (let i = blocks.length - 1; i >= 0; i--) {
 		if (blocks[i].type === "text") {
-			blocks[i] = { ...blocks[i], cache_control: cacheControl };
-			return;
+			if (blocks[i].cache_control != null) return false;
+			blocks[i] = { ...blocks[i], cache_control: cloneAnthropicCacheControl(cacheControl) };
+			return true;
 		}
 	}
-	applyCacheControlToLastBlock(blocks, cacheControl);
+	return false;
 }
 
 function applyPromptCaching(params: MessageCreateParamsStreaming, cacheControl?: AnthropicCacheControl): void {
 	if (!cacheControl) return;
 
-	// Skip if cache_control breakpoints were already placed externally on messages.
-	for (const message of params.messages) {
-		if (Array.isArray(message.content)) {
-			if ((message.content as Array<ContentBlockParam & CacheControlBlock>).some(b => b.cache_control != null))
-				return;
-		}
-	}
-
 	const MAX_CACHE_BREAKPOINTS = 4;
-	let cacheBreakpointsUsed = 0;
+	let cacheBreakpointsUsed = countCacheControlBreakpoints(params);
+	if (cacheBreakpointsUsed >= MAX_CACHE_BREAKPOINTS) return;
 	let isCCLayout = false;
 
 	if (params.system && Array.isArray(params.system) && params.system.length > 0) {
@@ -2122,9 +2228,12 @@ function applyPromptCaching(params: MessageCreateParamsStreaming, cacheControl?:
 			params.system.length >= 3 &&
 			(params.system[0] as { text?: string }).text?.startsWith(CLAUDE_BILLING_HEADER_PREFIX) === true;
 		if (isCCLayout) {
-			cacheBreakpointsUsed += applyClaudeCodeSystemCache(params.system as AnthropicSystemBlock[], cacheControl);
-		} else {
-			applyCacheControlToLastBlock(params.system, cacheControl);
+			const placed = Math.min(
+				MAX_CACHE_BREAKPOINTS - cacheBreakpointsUsed,
+				applyClaudeCodeSystemCache(params.system as AnthropicSystemBlock[], cacheControl),
+			);
+			cacheBreakpointsUsed += placed;
+		} else if (applyCacheControlToLastBlock(params.system, cacheControl)) {
 			cacheBreakpointsUsed++;
 		}
 	}
@@ -2137,14 +2246,19 @@ function applyPromptCaching(params: MessageCreateParamsStreaming, cacheControl?:
 		const message = params.messages[i];
 		if (!message) continue;
 		if (typeof message.content === "string") {
-			message.content = [{ type: "text", text: message.content, cache_control: cacheControl }];
+			message.content = [
+				{ type: "text", text: message.content, cache_control: cloneAnthropicCacheControl(cacheControl) },
+			];
 			cacheBreakpointsUsed++;
 		} else if (Array.isArray(message.content) && message.content.length > 0) {
-			applyCacheControlToLastTextBlock(
-				message.content as Array<ContentBlockParam & CacheControlBlock>,
-				cacheControl,
-			);
-			cacheBreakpointsUsed++;
+			if (
+				applyCacheControlToLastTextBlock(
+					message.content as Array<ContentBlockParam & CacheControlBlock>,
+					cacheControl,
+				)
+			) {
+				cacheBreakpointsUsed++;
+			}
 		}
 	}
 }
@@ -2157,7 +2271,9 @@ function normalizeCacheControlBlockTtl(block: CacheControlBlock, seenFiveMinute:
 		return;
 	}
 	if (seenFiveMinute.value) {
-		delete cacheControl.ttl;
+		const normalized = cloneAnthropicCacheControl(cacheControl);
+		delete normalized.ttl;
+		block.cache_control = normalized;
 	}
 }
 
@@ -2322,7 +2438,7 @@ function buildParams(
 	});
 
 	// Pre-compute tools.
-	let tools: ReturnType<typeof convertTools> | undefined;
+	let tools: AnthropicWireTool[] | undefined;
 	if (context.tools) {
 		tools = convertTools(
 			context.tools,
@@ -2385,11 +2501,11 @@ function buildParams(
 	// metadata → max_tokens → thinking → context_management → output_config → stream.
 	const params: MessageCreateParamsStreaming = {
 		model: model.id,
-		messages: convertAnthropicMessages(context.messages, model, isOAuthToken),
+		messages: convertAnthropicMessages(context.messages, model, isOAuthToken, baseUrl),
 		...(systemBlocks && { system: systemBlocks }),
 		...(tools !== undefined && { tools }),
 		...(metadata && { metadata }),
-		max_tokens: Math.min(CLAUDE_CODE_MAX_OUTPUT_TOKENS, options?.maxTokens || model.maxTokens),
+		max_tokens: Math.min(CLAUDE_CODE_MAX_OUTPUT_TOKENS, model.maxTokens, options?.maxTokens || model.maxTokens),
 		...(thinking && { thinking }),
 		...(contextManagement && { context_management: contextManagement }),
 		...(outputConfig && { output_config: outputConfig }),
@@ -2397,8 +2513,8 @@ function buildParams(
 	};
 
 	// Opus 4.7+ rejects non-default sampling parameters with 400 error.
-	const allowSamplingParams = !hasOpus47ApiRestrictions(model.id);
-	if (allowSamplingParams && options?.temperature !== undefined && !options?.thinkingEnabled) {
+	const allowSamplingParams = !hasOpus47ApiRestrictions(model.id) && !params.thinking;
+	if (allowSamplingParams && options?.temperature !== undefined) {
 		params.temperature = options.temperature;
 	}
 	if (allowSamplingParams && options?.topP !== undefined) {
@@ -2476,9 +2592,8 @@ function isZaiAnthropicEndpoint(model: Model<"anthropic-messages">): boolean {
  * arguments (#2005). Known non-signing hosts are also preserved for
  * compatibility.
  */
-function shouldReplayUnsignedThinking(model: Model<"anthropic-messages">): boolean {
+function shouldReplayUnsignedThinking(model: Model<"anthropic-messages">, baseUrl: string | undefined): boolean {
 	if (model.provider === "zai" || model.provider === "deepseek") return true;
-	const baseUrl = model.baseUrl;
 	if (baseUrl) {
 		try {
 			const hostname = new URL(baseUrl).hostname.toLowerCase();
@@ -2514,12 +2629,13 @@ export function convertAnthropicMessages(
 	messages: Message[],
 	model: Model<"anthropic-messages">,
 	isOAuthToken: boolean,
+	baseUrl = resolveAnthropicBaseUrl(model),
 ): AnthropicMessageParam[] {
-	const params: AnthropicMessageParam[] = [];
 	// Indices of params emitted from `developer` messages. After the main pass,
 	// the ones whose placement satisfies Anthropic's mid-conversation rules are
 	// upgraded from the `user` role to the authoritative `system` role.
 	const developerParamIndices: number[] = [];
+	const params: AnthropicMessageParam[] = [];
 
 	const transformedMessages = transformMessages(messages, model, normalizeToolCallId);
 
@@ -2578,7 +2694,7 @@ export function convertAnthropicMessages(
 					}
 					if (block.thinking.trim().length === 0) continue;
 					if (!block.thinkingSignature || block.thinkingSignature.trim().length === 0) {
-						if (shouldReplayUnsignedThinking(model)) {
+						if (shouldReplayUnsignedThinking(model, baseUrl)) {
 							blocks.push({
 								type: "thinking",
 								thinking: block.thinking.toWellFormed(),
@@ -2728,6 +2844,7 @@ function isJsonSchemaArrayNode(schema: Record<string, unknown>): boolean {
 	const t = schema.type;
 	if (t === "array") return true;
 	if (Array.isArray(t) && t.includes("array") && !t.includes("object")) return true;
+	if (schema.items !== undefined || Array.isArray(schema.prefixItems)) return true;
 	return false;
 }
 
@@ -2754,6 +2871,14 @@ function pickAnthropicScalarType(type: unknown): string | undefined {
 	}
 	return undefined;
 }
+function pickAnthropicEffectiveScalarType(schema: Record<string, unknown>): string | undefined {
+	const explicit = pickAnthropicScalarType(schema.type);
+	if (explicit) return explicit;
+	if (isRecord(schema.properties)) return "object";
+	if (schema.items !== undefined || Array.isArray(schema.prefixItems)) return "array";
+	return undefined;
+}
+
 
 function anthropicPerTypeKeep(scalarType: string | undefined): Set<string> | undefined {
 	switch (scalarType) {
@@ -2768,14 +2893,6 @@ function anthropicPerTypeKeep(scalarType: string | undefined): Set<string> | und
 	}
 }
 
-/**
- * Per-schema-object memoization slot for the normalized Anthropic tool form. We stamp
- * the result onto the host via a `Symbol` property (mirroring `utils/schema/stamps.ts`)
- * instead of using a `WeakMap`: it's a single hidden-class slot, so warm reads are
- * direct property access and write-once cycles resolve to the in-progress result.
- */
-const kAnthropicToolNormal = Symbol("pi.schema.anthropic.toolNormal");
-
 /**
  * Normalize a JSON Schema node for Anthropic tool `input_schema`.
  *
@@ -2796,20 +2913,20 @@ const kAnthropicToolNormal = Symbol("pi.schema.anthropic.toolNormal");
  * pass downstream demotes those shapes to non-strict instead of fabricating a closed
  * object, so callers like the resolve tool keep working open-map semantics.
  */
-export function normalizeAnthropicToolSchema(schema: unknown): unknown {
-	if (Array.isArray(schema)) return schema.map(entry => normalizeAnthropicToolSchema(entry));
+function normalizeAnthropicToolSchemaNode(
+	schema: unknown,
+	cache: WeakMap<Record<string, unknown>, Record<string, unknown>>,
+): unknown {
+	if (Array.isArray(schema)) return schema.map(entry => normalizeAnthropicToolSchemaNode(entry, cache));
 	if (!isRecord(schema)) return schema;
 
-	const slot = schema as Record<symbol, Record<string, unknown> | undefined>;
-	const existing = slot[kAnthropicToolNormal];
+	const existing = cache.get(schema);
 	if (existing !== undefined) return existing;
 
 	const result: Record<string, unknown> = {};
-	// Pre-stamp before recursion so cyclic schemas resolve to the in-progress object
-	// (mirrors the WeakMap-set-before-recurse pattern the original implementation used).
-	Object.defineProperty(schema, kAnthropicToolNormal, { value: result, writable: true, configurable: true });
+	cache.set(schema, result);
 
-	const scalarType = pickAnthropicScalarType(schema.type);
+	const scalarType = pickAnthropicEffectiveScalarType(schema);
 	const perTypeKeep = anthropicPerTypeKeep(scalarType);
 	const spill: Array<[string, unknown]> = [];
 
@@ -2848,12 +2965,12 @@ export function normalizeAnthropicToolSchema(schema: unknown): unknown {
 		const sourceProperties = result.properties as Record<string, unknown>;
 		for (const propName in sourceProperties) {
 			if (!Object.hasOwn(sourceProperties, propName)) continue;
-			normalizedProperties[propName] = normalizeAnthropicToolSchema(sourceProperties[propName]);
+			normalizedProperties[propName] = normalizeAnthropicToolSchemaNode(sourceProperties[propName], cache);
 		}
 		result.properties = normalizedProperties;
 	}
 	if (isRecord(result.additionalProperties)) {
-		const normalized = normalizeAnthropicToolSchema(result.additionalProperties);
+		const normalized = normalizeAnthropicToolSchemaNode(result.additionalProperties, cache);
 		if (isRecord(normalized) && Object.keys(normalized).length === 0) {
 			result.additionalProperties = true;
 		} else {
@@ -2861,17 +2978,17 @@ export function normalizeAnthropicToolSchema(schema: unknown): unknown {
 		}
 	}
 	if (Array.isArray(result.items)) {
-		result.items = result.items.map(item => normalizeAnthropicToolSchema(item));
+		result.items = result.items.map(item => normalizeAnthropicToolSchemaNode(item, cache));
 	} else if (isRecord(result.items)) {
-		result.items = normalizeAnthropicToolSchema(result.items);
+		result.items = normalizeAnthropicToolSchemaNode(result.items, cache);
 	}
 	if (Array.isArray(result.prefixItems)) {
-		result.prefixItems = result.prefixItems.map(item => normalizeAnthropicToolSchema(item));
+		result.prefixItems = result.prefixItems.map(item => normalizeAnthropicToolSchemaNode(item, cache));
 	}
 	for (const key of COMBINATOR_KEYS) {
 		const variants = result[key];
 		if (Array.isArray(variants)) {
-			result[key] = variants.map(variant => normalizeAnthropicToolSchema(variant));
+			result[key] = variants.map(variant => normalizeAnthropicToolSchemaNode(variant, cache));
 		}
 	}
 	for (const defsKey of ["$defs", "definitions"] as const) {
@@ -2881,7 +2998,7 @@ export function normalizeAnthropicToolSchema(schema: unknown): unknown {
 		const sourceDefs = definitions as Record<string, unknown>;
 		for (const name in sourceDefs) {
 			if (!Object.hasOwn(sourceDefs, name)) continue;
-			normalizedDefs[name] = normalizeAnthropicToolSchema(sourceDefs[name]);
+			normalizedDefs[name] = normalizeAnthropicToolSchemaNode(sourceDefs[name], cache);
 		}
 		result[defsKey] = normalizedDefs;
 	}
@@ -2890,6 +3007,10 @@ export function normalizeAnthropicToolSchema(schema: unknown): unknown {
 	return result;
 }
 
+export function normalizeAnthropicToolSchema(schema: unknown): unknown {
+	return normalizeAnthropicToolSchemaNode(schema, new WeakMap());
+}
+
 type AnthropicToolSchemaPlan = {
 	inputSchema: AnthropicToolInputSchema;
 	strict: boolean;
@@ -2910,6 +3031,25 @@ function hasNullVariant(schema: Record<string, unknown>): boolean {
 	if (Array.isArray(schema.type) && schema.type.includes("null")) return true;
 	return Array.isArray(schema.anyOf) && schema.anyOf.some(variant => isRecord(variant) && variant.type === "null");
 }
+function hasAnthropicSchemaDefiningKeyword(schema: Record<string, unknown>): boolean {
+	if (
+		schema.type !== undefined ||
+		schema.properties !== undefined ||
+		schema.additionalProperties !== undefined ||
+		schema.items !== undefined ||
+		schema.prefixItems !== undefined ||
+		schema.enum !== undefined ||
+		schema.const !== undefined ||
+		schema.$ref !== undefined
+	) {
+		return true;
+	}
+	for (const key of COMBINATOR_KEYS) {
+		if (schema[key] !== undefined) return true;
+	}
+	return schema.$defs !== undefined || schema.definitions !== undefined;
+}
+
 
 function makeAnthropicNullableSchema(schema: unknown, budget: AnthropicStrictBudget): unknown | undefined {
 	if (isRecord(schema)) {
@@ -2948,6 +3088,8 @@ function normalizeAnthropicStrictSchemaNode(
 	const cached = cache.get(schema);
 	if (cached) return cached;
 
+	if (!hasAnthropicSchemaDefiningKeyword(schema)) return undefined;
+
 	// Strict tool use only supports closed objects. Open maps stay available on
 	// the non-strict schema plan instead of producing an Anthropic 400.
 	if (isJsonSchemaObjectNode(schema) && schema.additionalProperties !== false) {
diff --git a/packages/ai/test/anthropic-stream-envelope.test.ts b/packages/ai/test/anthropic-stream-envelope.test.ts
index 42ea9a377..a10fc8df6 100644
--- a/packages/ai/test/anthropic-stream-envelope.test.ts
+++ b/packages/ai/test/anthropic-stream-envelope.test.ts
@@ -457,11 +457,11 @@ describe("anthropic stream envelope handling", () => {
 		expect(attempt).toBe(1);
 		expect(countEvents(events, "toolcall_start")).toBe(1);
 		expect(countEvents(events, "toolcall_delta")).toBe(1);
-		expect(countEvents(events, "toolcall_end")).toBe(1);
+		expect(countEvents(events, "toolcall_end")).toBe(0);
 		expect(countEvents(events, "error")).toBe(1);
 		expect(countEvents(events, "done")).toBe(0);
 		expect(result.stopReason).toBe("error");
-		expect(result.errorMessage).toContain("stream ended before terminal stop signal");
+		expect(result.errorMessage).toContain("Unterminated string");
 
 		const toolCall = result.content[0];
 		expect(toolCall?.type).toBe("toolCall");
@@ -494,7 +494,7 @@ describe("anthropic stream envelope handling", () => {
 		expect(countEvents(events, "done")).toBe(0);
 		expect(countEvents(events, "error")).toBe(1);
 		expect(result.stopReason).toBe("error");
-		expect(result.errorMessage).toContain("unterminated tool_use block");
+		expect(result.errorMessage).toContain("unterminated toolCall block");
 	});
 	it("parses raw SSE directly so unknown events do not fail Anthropic streams", async () => {
 		vi.spyOn(AnthropicMessages.prototype, "create").mockImplementation(
@@ -541,7 +541,7 @@ describe("anthropic stream envelope handling", () => {
 		expect(result.content).toEqual([{ type: "text", text: "partial" }]);
 	});
 
-	it("repairs malformed JSON in raw SSE event data before parsing", async () => {
+	it("surfaces malformed raw SSE event JSON instead of repairing protocol frames", async () => {
 		const malformedTextDelta =
 			'{"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"line\\qbreak"}}';
 		const successEvents = createTextSuccessEvents("unused");
@@ -556,13 +556,16 @@ describe("anthropic stream envelope handling", () => {
 		vi.spyOn(AnthropicMessages.prototype, "create").mockImplementation(() => createRawSseRequest(frames) as never);
 
 		const stream = streamAnthropic(model, context, { apiKey: "sk-ant-test" });
-		for await (const _ of stream) {
-			// drain stream
+		const events: AssistantMessageEvent[] = [];
+		for await (const event of stream) {
+			events.push(event);
 		}
 		const result = await stream.result();
 
-		expect(result.stopReason).toBe("stop");
-		expect(result.content).toEqual([{ type: "text", text: "line\\qbreak" }]);
+		expect(countEvents(events, "error")).toBe(1);
+		expect(countEvents(events, "done")).toBe(0);
+		expect(result.stopReason).toBe("error");
+		expect(result.errorMessage).toContain("Could not parse Anthropic SSE event content_block_delta");
 	});
 	it("surfaces a refusal fallback message when stop_details is null", async () => {
 		const refusalEvents: MockAnthropicEvent[] = [
diff --git a/packages/coding-agent/CHANGELOG.md b/packages/coding-agent/CHANGELOG.md
index a3ba00a7d..3411f6528 100644
--- a/packages/coding-agent/CHANGELOG.md
+++ b/packages/coding-agent/CHANGELOG.md
@@ -1,6 +1,7 @@
 # Changelog
 
 ## [Unreleased]
+
 ### Added
 
 - Added support for paste marker highlighting with accent styling (`[Paste #N, +X lines]`/`[Paste #N, Y chars]`) in the prompt editor, matching the visual treatment of image references
@@ -8,10 +9,15 @@
 
 ### Changed
 
+- Normalized image content before it enters model context so attached images are downscaled and preprocessed for prompts, steering messages, follow-ups, and custom agent messages
 - Changed image marker format to include pixel dimensions when available (`[Image #N, WxH]`), falling back to bare `[Image #N]` when header cannot be decoded
 - Changed the prompt editor to highlight large-paste placeholders (`[Paste #N, +X lines]`/`[Paste #N, Y chars]`) with the same accent styling as image references (bold, no hyperlink), and to delete image/paste markers atomically: a single backspace or forward-delete removes the whole marker instead of leaving a broken `[Paste #N, +X lines` behind.
 - Browser tool helpers (`tab.*`) are now individually tracked and time-bounded: when a `run` cell hits its budget, the timeout error names the still-running helper(s) and how long each has been stalled (e.g. `... (stalled on tab.screenshot({ selector: ".x" }) (29.9s))`) instead of the opaque `Browser code execution timed out after 30000ms`. Page-coupled helpers that should resolve quickly (`observe`, `screenshot`, `extract`) also fail fast with a named per-op error at `min(cellBudget, 20s)`, leaving budget for the rest of the cell, rather than silently consuming the whole budget.
 
+### Removed
+
+- Removed the special Anthropic `claude-opus-4-8` tool-call batch cap; sessions no longer abort an in-flight provider stream after a fixed number of completed tool calls.
+
 ### Fixed
 
 - Fixed task-row shimmer timing so every running description starts its highlight on the first character together and reaches the last character together, regardless of text length.
@@ -23,10 +29,6 @@
 - Fixed edit tool result previews to show only current-file lines and collapse long inserted blocks instead of echoing removed content.
 - Fixed `generateDiffString` to omit the mid-skip `...` placeholder between two nearby edits, conveying the elided gap via the jump in line numbers instead (consistent with how leading/trailing context skips already render). The placeholder row was indistinguishable from a genuine `...` context line and wasted a row in compact previews.
 
-### Removed
-
-- Removed the special Anthropic `claude-opus-4-8` tool-call batch cap; sessions no longer abort an in-flight provider stream after a fixed number of completed tool calls.
-
 ## [15.10.4] - 2026-06-08
 
 ### Added
diff --git a/packages/coding-agent/src/session/agent-session.ts b/packages/coding-agent/src/session/agent-session.ts
index 7d56f4228..f4ea08104 100644
--- a/packages/coding-agent/src/session/agent-session.ts
+++ b/packages/coding-agent/src/session/agent-session.ts
@@ -215,6 +215,7 @@ import { parseCommandArgs } from "../utils/command-args";
 import { type EditMode, resolveEditMode } from "../utils/edit-mode";
 import { resolveFileDisplayMode } from "../utils/file-display-mode";
 import { extractFileMentions, generateFileMentionMessages } from "../utils/file-mentions";
+import { normalizeModelContextImages } from "../utils/image-loading";
 import { buildNamedToolChoice } from "../utils/tool-choice";
 import type { AuthStorage } from "./auth-storage";
 import type { ClientBridge, ClientBridgePermissionOption, ClientBridgePermissionOutcome } from "./client-bridge";
@@ -4272,6 +4273,24 @@ export class AgentSession {
 		};
 	}
 
+	async #normalizeMessageContentImages(
+		content: string | (TextContent | ImageContent)[],
+	): Promise<string | (TextContent | ImageContent)[]> {
+		if (typeof content === "string") return content;
+		const images = content.filter((part): part is ImageContent => part.type === "image");
+		if (images.length === 0) return content;
+		const normalizedImages = await normalizeModelContextImages(images);
+		if (!normalizedImages) return content;
+		let imageIndex = 0;
+		return content.map(part => (part.type === "image" ? normalizedImages[imageIndex++]! : part));
+	}
+
+	async #normalizeAgentMessageImages<T extends AgentMessage>(message: T): Promise<T> {
+		const content = await this.#normalizeMessageContentImages(message.content);
+		if (content === message.content) return message;
+		return { ...message, content } as T;
+	}
+
 	/**
 	 * Send a prompt to the agent.
 	 * - Handles extension commands (registered via pi.registerCommand) immediately, even during streaming
@@ -4371,10 +4390,11 @@ export class AgentSession {
 		const hasPendingUserDirective = this.#toolChoiceQueue.inspect().includes("user-force");
 		const eagerTodoPrelude =
 			!options?.synthetic && !hasPendingUserDirective ? this.#createEagerTodoPrelude(expandedText) : undefined;
+		const normalizedImages = await normalizeModelContextImages(options?.images);
 
 		const userContent: (TextContent | ImageContent)[] = [{ type: "text", text: expandedText }];
-		if (options?.images) {
-			userContent.push(...options.images);
+		if (normalizedImages) {
+			userContent.push(...normalizedImages);
 		}
 
 		const promptAttribution = options?.attribution ?? (options?.synthetic ? "agent" : "user");
@@ -4391,6 +4411,7 @@ export class AgentSession {
 		try {
 			await this.#promptWithMessage(message, expandedText, {
 				...options,
+				images: normalizedImages,
 				prependMessages: eagerTodoPrelude ? [eagerTodoPrelude.message] : undefined,
 				appendMessages: keywordNotices.length > 0 ? keywordNotices : undefined,
 			});
@@ -4533,7 +4554,9 @@ export class AgentSession {
 					useHashLines: resolveFileDisplayMode(this).hashLines,
 					snapshotStore: getFileSnapshotStore(this),
 				});
-				messages.push(...fileMentionMessages);
+				for (const fileMentionMessage of fileMentionMessages) {
+					messages.push(await this.#normalizeAgentMessageImages(fileMentionMessage));
+				}
 			}
 
 			const beforeAgentStartSystemPrompt = await this.#buildSystemPromptForAgentStart(expandedText);
@@ -4549,15 +4572,17 @@ export class AgentSession {
 					const promptAttribution: "user" | "agent" | undefined =
 						"attribution" in message ? message.attribution : undefined;
 					for (const msg of result.messages) {
-						messages.push({
-							role: "custom",
-							customType: msg.customType,
-							content: msg.content,
-							display: msg.display,
-							details: msg.details,
-							attribution: msg.attribution ?? promptAttribution ?? (message.role === "user" ? "user" : "agent"),
-							timestamp: Date.now(),
-						});
+						messages.push(
+							await this.#normalizeAgentMessageImages({
+								role: "custom",
+								customType: msg.customType,
+								content: msg.content,
+								display: msg.display,
+								details: msg.details,
+								attribution: msg.attribution ?? promptAttribution ?? (message.role === "user" ? "user" : "agent"),
+								timestamp: Date.now(),
+							}),
+						);
 					}
 				}
 
@@ -4765,11 +4790,12 @@ export class AgentSession {
 	 * Internal: Queue a steering message (already expanded, no extension command check).
 	 */
 	async #queueSteer(text: string, images?: ImageContent[]): Promise<void> {
+		const normalizedImages = await normalizeModelContextImages(images);
 		const displayText = text || (images && images.length > 0 ? "[Image]" : "");
 		this.#steeringMessages.push({ text: displayText });
 		const content: (TextContent | ImageContent)[] = [{ type: "text", text }];
-		if (images && images.length > 0) {
-			content.push(...images);
+		if (normalizedImages && normalizedImages.length > 0) {
+			content.push(...normalizedImages);
 		}
 		this.agent.steer({
 			role: "user",
@@ -4784,11 +4810,12 @@ export class AgentSession {
 	 * Internal: Queue a follow-up message (already expanded, no extension command check).
 	 */
 	async #queueFollowUp(text: string, images?: ImageContent[]): Promise<void> {
+		const normalizedImages = await normalizeModelContextImages(images);
 		const displayText = text || (images && images.length > 0 ? "[Image]" : "");
 		this.#followUpMessages.push({ text: displayText });
 		const content: (TextContent | ImageContent)[] = [{ type: "text", text }];
-		if (images && images.length > 0) {
-			content.push(...images);
+		if (normalizedImages && normalizedImages.length > 0) {
+			content.push(...normalizedImages);
 		}
 		this.agent.followUp({
 			role: "user",
@@ -4932,16 +4959,17 @@ export class AgentSession {
 			attribution: message.attribution ?? "agent",
 			timestamp: Date.now(),
 		};
+		const normalizedAppMessage = await this.#normalizeAgentMessageImages(appMessage);
 		if (this.isStreaming) {
 			if (options?.deliverAs === "nextTurn") {
-				this.#queueHiddenNextTurnMessage(appMessage, options?.triggerTurn ?? false);
+				this.#queueHiddenNextTurnMessage(normalizedAppMessage, options?.triggerTurn ?? false);
 				return;
 			}
 
 			if (options?.deliverAs === "followUp") {
-				this.agent.followUp(appMessage);
+				this.agent.followUp(normalizedAppMessage);
 			} else {
-				this.agent.steer(appMessage);
+				this.agent.steer(normalizedAppMessage);
 			}
 			return;
 		}
@@ -4949,16 +4977,16 @@ export class AgentSession {
 		if (options?.deliverAs === "nextTurn") {
 			if (options?.triggerTurn) {
 				if (this.#clientBridge?.deferAgentInitiatedTurns && !this.#allowAcpAgentInitiatedTurns) {
-					this.#queueHiddenNextTurnMessage(appMessage, false);
+					this.#queueHiddenNextTurnMessage(normalizedAppMessage, false);
 					return;
 				}
-				await this.agent.prompt(appMessage);
+				await this.agent.prompt(normalizedAppMessage);
 				return;
 			}
-			this.agent.appendMessage(appMessage);
+			this.agent.appendMessage(normalizedAppMessage);
 			this.sessionManager.appendCustomMessageEntry(
-				message.customType,
-				message.content,
+				normalizedAppMessage.customType,
+				normalizedAppMessage.content,
 				message.display,
 				message.details,
 				message.attribution ?? "agent",
@@ -4968,17 +4996,17 @@ export class AgentSession {
 
 		if (options?.triggerTurn) {
 			if (this.#clientBridge?.deferAgentInitiatedTurns && !this.#allowAcpAgentInitiatedTurns) {
-				this.#queueHiddenNextTurnMessage(appMessage, false);
+				this.#queueHiddenNextTurnMessage(normalizedAppMessage, false);
 				return;
 			}
-			await this.agent.prompt(appMessage);
+			await this.agent.prompt(normalizedAppMessage);
 			return;
 		}
 
-		this.agent.appendMessage(appMessage);
+		this.agent.appendMessage(normalizedAppMessage);
 		this.sessionManager.appendCustomMessageEntry(
-			message.customType,
-			message.content,
+			normalizedAppMessage.customType,
+			normalizedAppMessage.content,
 			message.display,
 			message.details,
 			message.attribution ?? "agent",
diff --git a/packages/coding-agent/src/utils/image-loading.ts b/packages/coding-agent/src/utils/image-loading.ts
index 9f172188d..4be0a534e 100644
--- a/packages/coding-agent/src/utils/image-loading.ts
+++ b/packages/coding-agent/src/utils/image-loading.ts
@@ -2,7 +2,7 @@ import * as fs from "node:fs/promises";
 import type { ImageContent } from "@oh-my-pi/pi-ai";
 import { formatBytes, readImageMetadata, SUPPORTED_IMAGE_MIME_TYPES } from "@oh-my-pi/pi-utils";
 import { resolveReadPath } from "../tools/path-utils";
-import { formatDimensionNote, resizeImage } from "./image-resize";
+import { formatDimensionNote, resizeImage, type ImageResizeOptions } from "./image-resize";
 
 export const MAX_IMAGE_INPUT_BYTES = 20 * 1024 * 1024;
 export const SUPPORTED_INPUT_IMAGE_MIME_TYPES = SUPPORTED_IMAGE_MIME_TYPES;
@@ -50,6 +50,36 @@ export async function ensureSupportedImageInput(image: ImageContent): Promise<Im
 	}
 }
 
+export interface NormalizeModelContextImagesOptions {
+	resize?: ImageResizeOptions;
+}
+
+/**
+ * Normalize image blocks before they enter agent/model context. This keeps
+ * provider request construction from having to resize an unbounded batch of
+ * large images on the streaming hot path. Images are processed sequentially on
+ * purpose: `resizeImage` may fan out multiple encoders for one image, so the
+ * outer image batch must stay bounded.
+ */
+export async function normalizeModelContextImages(
+	images: ImageContent[] | undefined,
+	options?: NormalizeModelContextImagesOptions,
+): Promise<ImageContent[] | undefined> {
+	if (!images || images.length === 0) return undefined;
+	const normalized: ImageContent[] = [];
+	for (const image of images) {
+		try {
+			const resized = await resizeImage(image, options?.resize);
+			normalized.push({ type: "image", data: resized.data, mimeType: resized.mimeType });
+		} catch {
+			// Preserve existing caller behavior for decode/resize failures: keep the
+			// user's image block rather than dropping it from the turn.
+			normalized.push(image);
+		}
+	}
+	return normalized;
+}
+
 export async function loadImageInput(options: LoadImageInputOptions): Promise<LoadedImageInput | null> {
 	const maxBytes = options.maxBytes ?? MAX_IMAGE_INPUT_BYTES;
 	const resolvedPath = options.resolvedPath ?? resolveReadPath(options.path, options.cwd);
diff --git a/packages/coding-agent/test/image-input-normalization.test.ts b/packages/coding-agent/test/image-input-normalization.test.ts
index 8e4a0ca2a..39bdddd99 100644
--- a/packages/coding-agent/test/image-input-normalization.test.ts
+++ b/packages/coding-agent/test/image-input-normalization.test.ts
@@ -1,5 +1,5 @@
 import { describe, expect, test } from "bun:test";
-import { ensureSupportedImageInput } from "../src/utils/image-loading";
+import { ensureSupportedImageInput, normalizeModelContextImages } from "../src/utils/image-loading";
 
 // 1x1 red PNG (69 bytes). Bun.Image sniffs format from bytes, so we can pass
 // this with a non-supported MIME type and the conversion path runs over the
@@ -7,6 +7,17 @@ import { ensureSupportedImageInput } from "../src/utils/image-loading";
 const RED_1X1_PNG_BASE64 =
 	"iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAIAAACQd1PeAAAADElEQVR4nGP4z8AAAAMBAQDJ/pLvAAAAAElFTkSuQmCC";
 
+async function makeRedPng(width: number, height: number): Promise<string> {
+	const seed = Buffer.from(RED_1X1_PNG_BASE64, "base64");
+	const upscaled = await new Bun.Image(seed).resize(width, height, { filter: "nearest" }).png().bytes();
+	return Buffer.from(upscaled).toBase64();
+}
+
+async function dimensions(image: { data: string }): Promise<{ width: number; height: number }> {
+	const metadata = await new Bun.Image(Buffer.from(image.data, "base64")).metadata();
+	return { width: metadata.width, height: metadata.height };
+}
+
 describe("ensureSupportedImageInput", () => {
 	test("passes supported mime types through unchanged", async () => {
 		const input = { type: "image" as const, data: RED_1X1_PNG_BASE64, mimeType: "image/png" };
@@ -41,3 +52,22 @@ describe("ensureSupportedImageInput", () => {
 		expect(result).toBeNull();
 	});
 });
+
+describe("normalizeModelContextImages", () => {
+	test("downscales multiple large images before model context", async () => {
+		const wide = { type: "image" as const, data: await makeRedPng(2000, 1500), mimeType: "image/png" };
+		const tall = { type: "image" as const, data: await makeRedPng(1200, 2200), mimeType: "image/png" };
+
+		const result = await normalizeModelContextImages([wide, tall]);
+
+		expect(result).toHaveLength(2);
+		expect(result?.[0]?.type).toBe("image");
+		expect(result?.[1]?.type).toBe("image");
+		const wideDims = await dimensions(result![0]!);
+		const tallDims = await dimensions(result![1]!);
+		expect(wideDims.width).toBeLessThanOrEqual(1568);
+		expect(wideDims.height).toBeLessThanOrEqual(1568);
+		expect(tallDims.width).toBeLessThanOrEqual(1568);
+		expect(tallDims.height).toBeLessThanOrEqual(1568);
+	});
+});