@oh-my-pi/pi-coding-agent 17.1.5 → 17.1.7

package/src/session/agent-session.ts

@@ -35,6 +35,8 @@ import {
 	type AgentTurnEndContext,
 	AppendOnlyContextManager,
 	type AsideMessage,
+	type BeforeToolCallContext,
+	type BeforeToolCallResult,
 	resolveTelemetry,
 	type StreamFn,
 	TERMINAL_TOOL_RESULT_ABORT_REASON,
@@ -70,6 +72,7 @@ import type {
 	ToolChoice,
 	ToolResultMessage,
 	UsageReport,
+	UserMessage,
 } from "@oh-my-pi/pi-ai";
 import { type Effort, streamSimple } from "@oh-my-pi/pi-ai";
 import * as AIError from "@oh-my-pi/pi-ai/error";
@@ -133,6 +136,7 @@ import type { CompactOptions, ContextUsage } from "../extensibility/extensions/t
 import type { HookCommandContext } from "../extensibility/hooks/types";
 import type { Skill, SkillWarning } from "../extensibility/skills";
 import { expandSlashCommand, type FileSlashCommand } from "../extensibility/slash-commands";
+import { normalizeToolEventInput, resolveToolEventInput } from "../extensibility/tool-event-input";
 import { GoalRuntime } from "../goals/runtime";
 import type { GoalModeState } from "../goals/state";
 import type { HindsightSessionState } from "../hindsight/state";
@@ -176,6 +180,7 @@ import {
 	toReasoningEffort,
 } from "../thinking";
 import { shutdownTinyTitleClient } from "../tiny/title-client";
+import { resolveApproval } from "../tools/approval";
 import { type AskToolDetails, type AskToolInput, recoverAskQuestions } from "../tools/ask";
 import { releaseTabsForOwner } from "../tools/browser/tab-supervisor";
 import type { CheckpointState, CompletedRewindState } from "../tools/checkpoint";
@@ -196,6 +201,7 @@ import type { EditMode } 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 type { InspectImageMode } from "../utils/inspect-image-mode";
 import { generateSessionTitle } from "../utils/title-generator";
 import { buildNamedToolChoice, isToolChoiceActive } from "../utils/tool-choice";
 import type { VibeModeState } from "../vibe/state";
@@ -429,6 +435,8 @@ export class AgentSession {
 	#scheduledHiddenNextTurnGeneration: number | undefined = undefined;
 	#queuedMessageDrainScheduled = false;
 	#planModeState: PlanModeState | undefined;
+	/** Session-scoped `/vision` override; undefined = follow persisted `inspect_image.mode`. */
+	#inspectImageModeOverride: InspectImageMode | undefined;
 	#vibeModeState: VibeModeState | undefined;
 	#goalModeState: GoalModeState | undefined;
 	#goalRuntime: GoalRuntime;
@@ -911,6 +919,7 @@ export class AgentSession {
 		this.#models = new ModelControls(modelControlsHost, {
 			scopedModels: config.scopedModels,
 			thinkingLevel: config.thinkingLevel,
+			thinkingLevelCeiling: config.thinkingLevelCeiling,
 			serviceTierByFamily: config.serviceTierByFamily,
 		});
 
@@ -928,6 +937,7 @@ export class AgentSession {
 			thinkingLevel: () => this.thinkingLevel,
 			configuredThinkingLevel: () => this.configuredThinkingLevel(),
 			setThinkingLevel: level => this.setThinkingLevel(level),
+			thinkingLevelCeiling: () => this.#models.thinkingLevelCeiling,
 			isDisposed: () => this.#isDisposed,
 			isStreaming: () => this.isStreaming,
 			isCompacting: () => this.isCompacting,
@@ -1079,20 +1089,24 @@ export class AgentSession {
 			emitNotice: (level, message, source) => this.emitNotice(level, message, source),
 			notifyCommandMetadataChanged: () => this.#notifyCommandMetadataChanged(),
 			localProtocolOptions: () => this.#localProtocolOptions(),
+			getInspectImageModeOverride: () => this.#inspectImageModeOverride,
+			setInspectImageModeOverride: mode => {
+				this.#inspectImageModeOverride = mode;
+			},
 		};
 		this.#tools = new SessionTools(sessionToolsHost, {
 			autoApprove: config.autoApprove,
 			toolRegistry: config.toolRegistry,
 			createVibeTools: config.createVibeTools,
 			createComputerTool: config.createComputerTool,
+			createInspectImageTool: config.createInspectImageTool,
 			builtInToolNames: config.builtInToolNames,
 			presentationPinnedToolNames: config.presentationPinnedToolNames,
 			ensureWriteRegistered: config.ensureWriteRegistered,
 			rebuildSystemPrompt: config.rebuildSystemPrompt,
 			getLocalCalendarDate: config.getLocalCalendarDate,
 			getMcpServerInstructions: config.getMcpServerInstructions,
-			xdevRegistry: config.xdevRegistry,
-			initialMountedXdevToolNames: config.initialMountedXdevToolNames,
+			xdev: config.xdev,
 			setActiveToolNames: config.setActiveToolNames,
 			baseSystemPrompt: this.agent.state.systemPrompt,
 			skills: config.skills,
@@ -1175,6 +1189,10 @@ export class AgentSession {
 		});
 		// Tool-result hook owns synchronous post-tool actions that must affect the current loop.
 		this.agent.afterToolCall = ctx => this.#afterToolCall(ctx);
+		// Pre-scheduling tool_call wiring: extension handlers run at arg-prep
+		// time so a block/revision lands before concurrency resolution,
+		// tool_execution_start, and the wrapper's approval gate.
+		this.agent.beforeToolCall = ctx => this.#beforeToolCall(ctx);
 		this.agent.providerSessionState = this.#providerSessionState;
 		this.#syncAgentSessionId();
 		this.#todo.syncFromBranch();
@@ -1471,7 +1489,7 @@ export class AgentSession {
 
 		this.#toolChoiceQueue.pushSequence([forced, "none"], {
 			label: "user-force",
-			onRejected: () => "requeue",
+			onRejected: info => (info.reason === "unavailable" ? "drop_sequence" : "requeue"),
 		});
 	}
 
@@ -2959,6 +2977,50 @@ export class AgentSession {
 		}
 		return this.#ttsr.afterToolCall(ctx);
 	}
+	/**
+	 * Emits the extension `tool_call` event for a loop-dispatched call at
+	 * arg-prep time — before concurrency scheduling, `tool_execution_start`,
+	 * and the wrapper's approval gate. A handler block becomes a blocked tool
+	 * result; a handler `input` revision becomes the arguments the loop
+	 * schedules, displays, persists, and executes, so approval resolves against
+	 * what actually runs. Marks the dispatch so `ExtensionToolWrapper` does not
+	 * emit a second event (nested xd:// device dispatches and direct non-loop
+	 * execution still emit there).
+	 */
+	async #beforeToolCall(ctx: BeforeToolCallContext): Promise<BeforeToolCallResult | undefined> {
+		const runner = this.#extensionRunner;
+		if (!runner?.hasHandlers("tool_call")) return undefined;
+		const metadata = ctx.toolCall.providerMetadata;
+		const computer = metadata?.type === "computer" ? metadata : undefined;
+		// Parity with the wrapper's pre-emit short-circuit: an already-denied
+		// call never reaches extensions. Deny is mode-independent (tool decision
+		// or user policy), so resolving under the most permissive mode is exact;
+		// the wrapper still enforces the mode-accurate gate before execution.
+		const userPolicies = (this.settings.get("tools.approval") ?? {}) as Record<string, unknown>;
+		const approvalArgs = computer ? { actions: computer.actions } : ctx.args;
+		if (resolveApproval(ctx.tool, approvalArgs, "yolo", userPolicies).policy === "deny") {
+			return undefined;
+		}
+		const eventArgs = computer
+			? { actions: computer.actions, pendingSafetyChecks: computer.pendingSafetyChecks }
+			: ctx.args;
+		runner.markToolCallEmitted(ctx.toolCall.id, ctx.tool.name);
+		const callResult = await runner.emitToolCall({
+			type: "tool_call",
+			toolName: ctx.tool.name,
+			toolCallId: ctx.toolCall.id,
+			input: normalizeToolEventInput(ctx.tool.name, resolveToolEventInput(ctx.tool, eventArgs)),
+		});
+		if (callResult?.block) {
+			return { block: true, reason: callResult.reason || "Tool execution was blocked by an extension" };
+		}
+		// A computer call's event input is a synthetic {actions, pendingSafetyChecks}
+		// view, not the execution params — a revision cannot map back onto them.
+		if (callResult?.input !== undefined && !computer) {
+			return { args: callResult.input };
+		}
+		return undefined;
+	}
 
 	/** Find the last assistant message in agent state (including aborted ones) */
 	#findLastAssistantMessage(): AssistantMessage | undefined {
@@ -3948,6 +4010,34 @@ export class AgentSession {
 		return this.#tools.setComputerToolEnabled(enabled);
 	}
 
+	/**
+	 * Session-scoped inspect_image mode (`/vision`). `auto` clears the override
+	 * and returns to the persisted `inspect_image.mode` setting; `on`/`off`
+	 * force the tool for this session only. See {@link SessionTools.setInspectImageMode}.
+	 */
+	setInspectImageMode(mode: InspectImageMode): Promise<boolean> {
+		return this.#tools.setInspectImageMode(mode);
+	}
+
+	/** Effective inspect_image state for `/vision status`. */
+	inspectImageState(): { mode: InspectImageMode; active: boolean; model: string | undefined } {
+		return this.#tools.inspectImageState();
+	}
+
+	/** Session-scoped `/vision` override; undefined means "follow the persisted setting". */
+	getInspectImageModeOverride(): InspectImageMode | undefined {
+		return this.#inspectImageModeOverride;
+	}
+
+	/**
+	 * Reconciles the inspect_image tool set after the persisted
+	 * `inspect_image.mode` setting changed (e.g. via the settings selector), so
+	 * the new value takes effect immediately instead of on the next model switch.
+	 */
+	applyInspectImageModeChange(): Promise<boolean> {
+		return this.#tools.reconcileInspectImageTool();
+	}
+
 	/** Cancels the local rollout-memory startup owned by this session. */
 	cancelLocalMemoryStartup(): void {
 		this.#memory.cancelLocalMemoryStartup();
@@ -4222,6 +4312,7 @@ export class AgentSession {
 
 	/** Drop mutable tool decisions and directives owned by the previous logical session. */
 	#clearSessionScopedToolState(): void {
+		this.agent.clearDeferredToolDirectives();
 		this.#toolChoiceQueue.clear();
 		this.#tools.clearAcpPermissionDecisions();
 	}
@@ -5901,6 +5992,10 @@ export class AgentSession {
 				additionalDirectories: this.settings.get("workspace.additionalDirectories"),
 			});
 			this.#bash.markSessionTransition(bashTransition);
+			// The new session owns the transcript from here, so the previous
+			// conversation's advisor spend is retired with it. Clearing at the commit
+			// point keeps the status line honest even if a later step below throws.
+			this.#advisors.clearCost();
 			sessionTransitioned = true;
 		} finally {
 			this.#bash.finishSessionTransition(bashTransition, sessionTransitioned);
@@ -6323,7 +6418,7 @@ export class AgentSession {
 			activeMessages.splice(0, activeMessages.length, ...sessionContext.messages);
 		}
 		this.agent.replaceMessages(activeMessages ?? sessionContext.messages);
-		this.#advisors.resetSessionState();
+		this.#advisors.resetSessionState({ preserveCost: true });
 		this.#todo.syncFromBranch();
 		this.#closeCodexProviderSessionsForHistoryRewrite();
 		this.#checkpointState = undefined;
@@ -6398,7 +6493,7 @@ export class AgentSession {
 		return true;
 	}
 
-	#setModelWithProviderSessionReset(model: Model): void {
+	async #setModelWithProviderSessionReset(model: Model): Promise<void> {
 		const currentModel = this.model;
 		if (currentModel) {
 			this.#closeProviderSessionsForModelSwitch(currentModel, model);
@@ -6410,6 +6505,16 @@ export class AgentSession {
 
 		// Re-evaluate append-only context mode — provider or setting may have changed
 		this.#syncAppendOnlyContext(model);
+
+		// inspect_image auto mode keys off model image capability. Reconcile
+		// centrally here so retry-fallback model changes (turn-recovery.ts),
+		// which bypass syncAfterModelChange, cannot leave the tool set stale —
+		// callers await, so a scheduled retry never races the reconciled slate.
+		try {
+			await this.#tools.reconcileInspectImageAfterModelChange();
+		} catch (error) {
+			logger.warn("inspect_image reconcile after model change failed", { error: String(error) });
+		}
 	}
 
 	#closeCodexProviderSessionsForHistoryRewrite(): void {
@@ -6941,7 +7046,7 @@ export class AgentSession {
 			}
 
 			this.agent.replaceMessages(sessionContext.messages);
-			this.#advisors.resetSessionState();
+			this.#advisors.resetSessionState({ preserveCost: true });
 			this.#todo.syncFromBranch();
 			if (switchingToDifferentSession) {
 				this.#closeAllProviderSessions("session switch");
@@ -6974,7 +7079,7 @@ export class AgentSession {
 								currentModel.id !== match.id ||
 								currentModel.api !== match.api));
 					if (shouldResetProviderState) {
-						this.#setModelWithProviderSessionReset(match);
+						await this.#setModelWithProviderSessionReset(match);
 					} else {
 						this.agent.setModel(match);
 					}
@@ -7050,6 +7155,9 @@ export class AgentSession {
 					error: String(refreshErr),
 				});
 			}
+			// Only a committed switch retires the previous conversation's advisor spend:
+			// an earlier clear would be lost work if any step above rolled the switch back.
+			if (switchingToDifferentSession) this.#advisors.clearCost();
 			this.#bash.finishSessionTransition(bashTransition, true);
 			return true;
 		} catch (error) {
@@ -7098,10 +7206,12 @@ export class AgentSession {
 	 * @param entryId ID of the entry to branch from
 	 * @returns Object with:
 	 *   - selectedText: The text of the selected user message (for editor pre-fill)
+	 *   - selectedImages: Image attachments of the selected user message (for editor draft restore)
 	 *   - cancelled: True if a hook cancelled the branch
 	 */
 	async branch(entryId: string): Promise<{
 		selectedText: string;
+		selectedImages: ImageContent[];
 		cancelled: boolean;
 	}> {
 		const previousSessionFile = this.sessionFile;
@@ -7112,6 +7222,7 @@ export class AgentSession {
 		}
 
 		const selectedText = this.#extractUserMessageText(selectedEntry.message.content);
+		const selectedImages = this.#extractUserMessageImages(selectedEntry.message.content);
 
 		let skipConversationRestore = false;
 
@@ -7123,7 +7234,7 @@ export class AgentSession {
 			})) as SessionBeforeBranchResult | undefined;
 
 			if (result?.cancel) {
-				return { selectedText, cancelled: true };
+				return { selectedText, selectedImages, cancelled: true };
 			}
 			skipConversationRestore = result?.skipConversationRestore ?? false;
 		}
@@ -7148,6 +7259,7 @@ export class AgentSession {
 				this.sessionManager.createBranchedSession(selectedEntry.parentId);
 			}
 			this.#bash.markSessionTransition(bashTransition);
+			this.#advisors.clearCost();
 			sessionTransitioned = true;
 		} finally {
 			this.#bash.finishSessionTransition(bashTransition, sessionTransitioned);
@@ -7178,7 +7290,7 @@ export class AgentSession {
 			this.#closeCodexProviderSessionsForHistoryRewrite();
 		}
 
-		return { selectedText, cancelled: false };
+		return { selectedText, selectedImages, cancelled: false };
 	}
 
 	async branchFromBtw(
@@ -7245,6 +7357,7 @@ export class AgentSession {
 		try {
 			this.sessionManager.createBranchedSession(leafId);
 			this.#bash.markSessionTransition(bashTransition);
+			this.#advisors.clearCost();
 			sessionTransitioned = true;
 		} finally {
 			this.#bash.finishSessionTransition(bashTransition, sessionTransitioned);
@@ -7292,7 +7405,7 @@ export class AgentSession {
 	 * @param targetId The entry ID to navigate to
 	 * @param options.summarize Whether user wants to summarize abandoned branch
 	 * @param options.customInstructions Custom instructions for summarizer
-	 * @returns Result with editorText (if user message) and cancelled status
+	 * @returns Result with editorText/editorImages (if user message) and cancelled status
 	 */
 	async navigateTree(
 		targetId: string,
@@ -7322,6 +7435,8 @@ export class AgentSession {
 		} = {},
 	): Promise<{
 		editorText?: string;
+		/** Image attachments of the target user message, parallel to the positional `[Image #N]` markers in {@link editorText}. */
+		editorImages?: ImageContent[];
 		cancelled: boolean;
 		aborted?: boolean;
 		summaryEntry?: BranchSummaryEntry;
@@ -7494,6 +7609,7 @@ export class AgentSession {
 		// Determine the new leaf position based on target type
 		let newLeafId: string | null;
 		let editorText: string | undefined;
+		let editorImages: ImageContent[] | undefined;
 		// Set when the second-pass `ask` re-answer branch below actually commits a
 		// new sibling answer — the trigger for resuming the agent afterwards so the
 		// model consumes it, mirroring a live `ask` completion (issue #6483).
@@ -7503,6 +7619,8 @@ export class AgentSession {
 			// User message: leaf = parent (null if root), text goes to editor
 			newLeafId = targetEntry.parentId;
 			editorText = this.#extractUserMessageText(targetEntry.message.content);
+			const targetImages = this.#extractUserMessageImages(targetEntry.message.content);
+			if (targetImages.length > 0) editorImages = targetImages;
 		} else if (targetEntry.type === "custom_message" && targetEntry.customType !== SKILL_PROMPT_MESSAGE_TYPE) {
 			// Custom message: leaf = parent (null if root), text goes to editor
 			newLeafId = targetEntry.parentId;
@@ -7574,7 +7692,7 @@ export class AgentSession {
 		const displayContext = deobfuscateSessionContext(stateContext, this.#obfuscator);
 		this.agent.replaceMessages(displayContext.messages);
 		this.#rehydrateCheckpointRewindState();
-		this.#advisors.resetSessionState();
+		this.#advisors.resetSessionState({ preserveCost: true });
 		this.#todo.syncFromBranch();
 		this.#closeCodexProviderSessionsForHistoryRewrite();
 
@@ -7602,6 +7720,7 @@ export class AgentSession {
 			const rawContext = this.sessionManager.buildSessionContext();
 			return {
 				editorText,
+				editorImages,
 				cancelled: false,
 				summaryEntry,
 				sessionContext: rawContext,
@@ -7610,6 +7729,7 @@ export class AgentSession {
 		}
 		return {
 			editorText,
+			editorImages,
 			cancelled: false,
 			summaryEntry,
 			sessionContext: stateContext,
@@ -7728,6 +7848,14 @@ export class AgentSession {
 		return "";
 	}
 
+	/** Image parts of a stored user message, in submission order — index N-1 backs the
+	 *  `[Image #N]` marker in the message text, so restoring them alongside the text keeps
+	 *  positional markers resolvable on resubmit. */
+	#extractUserMessageImages(content: UserMessage["content"]): ImageContent[] {
+		if (!Array.isArray(content)) return [];
+		return content.filter((c): c is ImageContent => c.type === "image");
+	}
+
 	/**
 	 * Get session statistics.
 	 */
@@ -8205,6 +8333,11 @@ export class AgentSession {
 	getAdvisorStatusOverview(): { configured: boolean; advisors: { name: string; status: AdvisorRuntimeStatus }[] } {
 		return this.#advisors.getAdvisorStatusOverview();
 	}
+
+	/** Return cumulative cost recorded for the current session's advisor activity. */
+	getAdvisorCost(): number {
+		return this.#advisors.getAdvisorCost();
+	}
 	/**
 	 * Return structured advisor stats for the status command and TUI panel.
 	 */

package/src/session/model-controls.ts

@@ -27,6 +27,7 @@ import {
 	AUTO_THINKING,
 	type ConfiguredThinkingLevel,
 	clampAutoThinkingEffort,
+	clampThinkingLevelToCeiling,
 	resolveProvisionalAutoLevel,
 	resolveThinkingLevelForModel,
 	shouldDisableReasoning,
@@ -51,7 +52,7 @@ export interface ModelControlsHost {
 	promptGeneration(): number;
 	resolveActiveEditMode(): EditMode;
 	syncAfterModelChange(previousEditMode: EditMode): Promise<void>;
-	setModelWithProviderSessionReset(model: Model): void;
+	setModelWithProviderSessionReset(model: Model): Promise<void>;
 	clearActiveRetryFallback(): void;
 	clearInheritedProviderPromptCacheKey(): void;
 	magicKeywordEnabled(keyword: "orchestrate" | "ultrathink" | "workflow"): boolean;
@@ -65,6 +66,8 @@ export class ModelControls {
 	readonly #host: ModelControlsHost;
 	#scopedModels: Array<{ model: Model; thinkingLevel?: ThinkingLevel }>;
 	#thinkingLevel: ThinkingLevel | undefined;
+	/** Hard per-session effort ceiling (e.g. a task spawn's `task.maxEffort` cap); recovery paths re-clamp to it. */
+	readonly #thinkingLevelCeiling: Effort | undefined;
 	#autoThinking = false;
 	#autoResolvedLevel: Effort | undefined;
 	#serviceTierByFamily: ServiceTierByFamily;
@@ -74,18 +77,28 @@ export class ModelControls {
 		options: {
 			scopedModels?: Array<{ model: Model; thinkingLevel?: ThinkingLevel }>;
 			thinkingLevel?: ConfiguredThinkingLevel;
+			thinkingLevelCeiling?: Effort;
 			serviceTierByFamily?: ServiceTierByFamily;
 		},
 	) {
 		this.#host = host;
 		this.#scopedModels = options.scopedModels ?? [];
 		this.#serviceTierByFamily = options.serviceTierByFamily ?? {};
+		this.#thinkingLevelCeiling = options.thinkingLevelCeiling;
 		if (options.thinkingLevel === AUTO_THINKING) {
 			// Keep auto pending until the first turn while exposing a valid wire effort.
 			this.#autoThinking = true;
-			this.#thinkingLevel = resolveProvisionalAutoLevel(this.#model);
+			this.#thinkingLevel = clampThinkingLevelToCeiling(
+				this.#model,
+				resolveProvisionalAutoLevel(this.#model),
+				this.#thinkingLevelCeiling,
+			);
 		} else {
-			this.#thinkingLevel = options.thinkingLevel;
+			this.#thinkingLevel = clampThinkingLevelToCeiling(
+				this.#model,
+				options.thinkingLevel,
+				this.#thinkingLevelCeiling,
+			);
 		}
 		this.#applyThinkingLevelToAgent(this.#thinkingLevel);
 	}
@@ -99,6 +112,11 @@ export class ModelControls {
 		return this.#thinkingLevel;
 	}
 
+	/** Hard per-session effort ceiling every thinking-level change is clamped to. */
+	get thinkingLevelCeiling(): Effort | undefined {
+		return this.#thinkingLevelCeiling;
+	}
+
 	/** Configured selector, preserving `auto` while classification is active. */
 	configuredThinkingLevel(): ConfiguredThinkingLevel | undefined {
 		return this.#autoThinking ? AUTO_THINKING : this.#thinkingLevel;
@@ -130,8 +148,15 @@ export class ModelControls {
 		this.#autoResolvedLevel = undefined;
 		this.#thinkingLevel =
 			level === AUTO_THINKING
-				? resolveProvisionalAutoLevel(this.#model)
-				: resolveThinkingLevelForModel(this.#model, level);
+				? clampThinkingLevelToCeiling(
+						this.#model,
+						resolveProvisionalAutoLevel(this.#model),
+						this.#thinkingLevelCeiling,
+					)
+				: resolveThinkingLevelForModel(
+						this.#model,
+						clampThinkingLevelToCeiling(this.#model, level, this.#thinkingLevelCeiling),
+					);
 		this.#applyThinkingLevelToAgent(this.#thinkingLevel);
 	}
 
@@ -195,7 +220,7 @@ export class ModelControls {
 
 		this.#host.modelRegistry.clearSuppressedSelector(formatModelStringWithRouting(targetModel));
 		this.#host.clearActiveRetryFallback();
-		this.#host.setModelWithProviderSessionReset(targetModel);
+		await this.#host.setModelWithProviderSessionReset(targetModel);
 		this.#host.sessionManager.appendModelChange(`${targetModel.provider}/${targetModel.id}`, role);
 		if (options?.persist) {
 			this.#host.settings.setModelRole(
@@ -240,7 +265,7 @@ export class ModelControls {
 
 		this.#host.modelRegistry.clearSuppressedSelector(formatModelStringWithRouting(targetModel));
 		this.#host.clearActiveRetryFallback();
-		this.#host.setModelWithProviderSessionReset(targetModel);
+		await this.#host.setModelWithProviderSessionReset(targetModel);
 		this.#host.sessionManager.appendModelChange(
 			`${targetModel.provider}/${targetModel.id}`,
 			options?.ephemeral ? EPHEMERAL_MODEL_CHANGE_ROLE : "temporary",
@@ -400,7 +425,7 @@ export class ModelControls {
 		// Apply model
 		this.#host.modelRegistry.clearSuppressedSelector(formatModelStringWithRouting(next.model));
 		this.#host.clearActiveRetryFallback();
-		this.#host.setModelWithProviderSessionReset(next.model);
+		await this.#host.setModelWithProviderSessionReset(next.model);
 		this.#host.sessionManager.appendModelChange(`${next.model.provider}/${next.model.id}`);
 		this.#host.settings.getStorage()?.recordModelUsage(`${next.model.provider}/${next.model.id}`);
 
@@ -431,7 +456,7 @@ export class ModelControls {
 
 		this.#host.modelRegistry.clearSuppressedSelector(formatModelStringWithRouting(nextModel));
 		this.#host.clearActiveRetryFallback();
-		this.#host.setModelWithProviderSessionReset(nextModel);
+		await this.#host.setModelWithProviderSessionReset(nextModel);
 		this.#host.sessionManager.appendModelChange(`${nextModel.provider}/${nextModel.id}`);
 		this.#host.settings.getStorage()?.recordModelUsage(`${nextModel.provider}/${nextModel.id}`);
 		// Re-apply the current thinking level (or auto) for the newly selected model
@@ -469,7 +494,11 @@ export class ModelControls {
 	 */
 	setThinkingLevel(level: ConfiguredThinkingLevel | undefined, persist: boolean = false): void {
 		if (level === AUTO_THINKING) {
-			const provisional = resolveProvisionalAutoLevel(this.#model);
+			const provisional = clampThinkingLevelToCeiling(
+				this.#model,
+				resolveProvisionalAutoLevel(this.#model),
+				this.#thinkingLevelCeiling,
+			);
 			const wasAuto = this.#autoThinking;
 			const previousLevel = this.#thinkingLevel;
 			this.#autoThinking = true;
@@ -493,7 +522,10 @@ export class ModelControls {
 		const wasAuto = this.#autoThinking;
 		this.#autoThinking = false;
 		this.#autoResolvedLevel = undefined;
-		const effectiveLevel = resolveThinkingLevelForModel(this.#model, level);
+		const effectiveLevel = resolveThinkingLevelForModel(
+			this.#model,
+			clampThinkingLevelToCeiling(this.#model, level, this.#thinkingLevelCeiling),
+		);
 		// Leaving auto must persist even when the resolved effort is unchanged (e.g.
 		// auto resolved to medium, then the user pins medium): otherwise the latest
 		// session entry keeps `configured: "auto"` and resume re-enables auto.
@@ -591,7 +623,11 @@ export class ModelControls {
 		// Drop the result if the turn was aborted/superseded while classifying.
 		if (this.#host.promptGeneration() !== generation || !this.#autoThinking) return;
 
-		const effort = resolved ?? resolveProvisionalAutoLevel(model);
+		const effort = clampThinkingLevelToCeiling(
+			model,
+			resolved ?? resolveProvisionalAutoLevel(model),
+			this.#thinkingLevelCeiling,
+		);
 		if (effort === undefined) return;
 		const shouldPersistResolution = this.#thinkingLevel !== effort;
 		this.#autoResolvedLevel = effort;