pi-subagents 0.24.2 → 0.24.4

package/src/runs/background/subagent-runner.ts

@@ -66,6 +66,7 @@ import {
 	formatWorktreeTaskCwdConflict,
 	type WorktreeSetup,
 } from "../shared/worktree.ts";
+import { resolveEffectiveThinking } from "../../shared/model-info.ts";
 import { writeInitialProgressFile } from "../../shared/settings.ts";
 
 interface SubagentRunConfig {
@@ -221,7 +222,12 @@ function runPiStreaming(
 			...(piPackageRoot ? { piPackageRoot } : {}),
 			...(piArgv1 ? { argv1: piArgv1 } : {}),
 		});
-		const child = spawn(spawnSpec.command, spawnSpec.args, { cwd, stdio: ["ignore", "pipe", "pipe"], env: spawnEnv });
+		const child = spawn(spawnSpec.command, spawnSpec.args, {
+			cwd,
+			stdio: ["ignore", "pipe", "pipe"],
+			env: spawnEnv,
+			windowsHide: true,
+		});
 		let stderr = "";
 		let stdoutBuf = "";
 		let stderrBuf = "";
@@ -229,6 +235,7 @@ function runPiStreaming(
 		const usage = emptyUsage();
 		let model: string | undefined;
 		let error: string | undefined;
+		let assistantError: string | undefined;
 		let interrupted = false;
 		let observedMutationAttempt = false;
 		const rawStdoutLines: string[] = [];
@@ -289,7 +296,7 @@ function runPiStreaming(
 
 				if (event.type !== "message_end" || event.message.role !== "assistant") return;
 				if (event.message.model) model = event.message.model;
-				if (event.message.errorMessage) error = event.message.errorMessage;
+				if (event.message.errorMessage) assistantError = event.message.errorMessage;
 				const eventUsage = event.message.usage;
 				if (eventUsage) {
 					usage.turns++;
@@ -303,6 +310,7 @@ function runPiStreaming(
 				const hasToolCall = Array.isArray(event.message.content)
 					&& event.message.content.some((part) => (part as { type?: string }).type === "toolCall");
 				if (stopReason === "stop" && !hasToolCall) {
+					if (!event.message.errorMessage && extractTextFromContent(event.message.content).trim()) assistantError = undefined;
 					cleanTerminalAssistantStopReceived ||= !event.message.errorMessage;
 					startFinalDrain();
 				}
@@ -370,7 +378,7 @@ function runPiStreaming(
 				const termSent = trySignalChild(child, "SIGTERM");
 				if (!termSent) return;
 				forcedTerminationSignal = true;
-				if (!cleanTerminalAssistantStopReceived && !error) {
+				if (!cleanTerminalAssistantStopReceived && !error && !assistantError) {
 					error = `Subagent process did not exit within ${FINAL_STOP_GRACE_MS}ms after its final message. Forcing termination.`;
 				}
 				finalHardKillTimer = setTimeout(() => {
@@ -394,14 +402,15 @@ function runPiStreaming(
 			if (stderrBuf.trim()) appendChildLine("subagent.child.stderr", stderrBuf);
 			outputStream.end();
 			const finalOutput = getFinalOutput(messages) || rawStdoutLines.join("\n").trim();
-			const forcedDrainAfterFinalSuccess = forcedTerminationSignal && cleanTerminalAssistantStopReceived && !error;
+			const finalError = error ?? assistantError;
+			const forcedDrainAfterFinalSuccess = forcedTerminationSignal && cleanTerminalAssistantStopReceived && !finalError;
 			resolve({
 				stderr,
 				exitCode: interrupted || forcedDrainAfterFinalSuccess ? 0 : forcedTerminationSignal || signal ? (exitCode ?? 1) : exitCode,
 				messages,
 				usage,
 				model,
-				error: interrupted || forcedDrainAfterFinalSuccess ? undefined : error,
+				error: interrupted || forcedDrainAfterFinalSuccess ? undefined : finalError,
 				finalOutput,
 				interrupted,
 				observedMutationAttempt,
@@ -416,7 +425,7 @@ function runPiStreaming(
 			outputStream.end();
 			const finalOutput = getFinalOutput(messages) || rawStdoutLines.join("\n").trim();
 			const spawnErrorMessage = spawnError instanceof Error ? spawnError.message : String(spawnError);
-			resolve({ stderr, exitCode: 1, messages, usage, model, error: error ?? spawnErrorMessage, finalOutput, observedMutationAttempt });
+			resolve({ stderr, exitCode: 1, messages, usage, model, error: error ?? assistantError ?? spawnErrorMessage, finalOutput, observedMutationAttempt });
 		});
 	});
 }
@@ -545,6 +554,7 @@ interface SingleStepContext {
 	registerInterrupt?: (interrupt: (() => void) | undefined) => void;
 	childIntercomTarget?: string;
 	orchestratorIntercomTarget?: string;
+	onAttemptStart?: (attempt: { model?: string; thinking?: string }) => void;
 	onChildEvent?: (event: ChildEvent) => void;
 }
 
@@ -596,6 +606,7 @@ async function runSingleStep(
 
 	for (let index = 0; index < candidates.length; index++) {
 		const candidate = candidates[index];
+		ctx.onAttemptStart?.({ model: candidate, thinking: resolveEffectiveThinking(candidate, step.thinking) });
 		const outputSnapshot = captureSingleOutputSnapshot(step.outputPath);
 		const { args, env, tempDir } = buildPiArgs({
 			baseArgs: ["--mode", "json", "-p"],
@@ -611,6 +622,7 @@ async function runSingleStep(
 			systemPrompt: step.systemPrompt,
 			systemPromptMode: step.systemPromptMode,
 			mcpDirectTools: step.mcpDirectTools,
+			cwd: step.cwd ?? ctx.cwd,
 			promptFileStem: step.agent,
 			intercomSessionName: ctx.childIntercomTarget,
 			orchestratorIntercomTarget: ctx.orchestratorIntercomTarget,
@@ -633,11 +645,13 @@ async function runSingleStep(
 		cleanupTempDir(tempDir);
 
 		const hiddenError = run.exitCode === 0 && !run.error ? detectSubagentError(run.messages) : null;
-		const completionGuard = run.exitCode === 0 && !run.error && !hiddenError?.hasError
+		const completionGuard = run.exitCode === 0 && !run.error && !hiddenError?.hasError && step.completionGuard !== false
 			? evaluateCompletionMutationGuard({
 				agent: step.agent,
 				task,
 				messages: run.messages,
+				tools: step.tools,
+				mcpDirectTools: step.mcpDirectTools,
 			})
 			: undefined;
 		const completionGuardTriggered = completionGuard?.triggered === true && !run.observedMutationAttempt;
@@ -912,6 +926,7 @@ async function runSubagent(config: SubagentRunConfig): Promise<void> {
 			...(step.sessionFile ? { sessionFile: step.sessionFile } : {}),
 			skills: step.skills,
 			model: step.model,
+			thinking: step.thinking,
 			attemptedModels: step.modelCandidates && step.modelCandidates.length > 0 ? step.modelCandidates : step.model ? [step.model] : undefined,
 			recentTools: [],
 			recentOutput: [],
@@ -1007,6 +1022,14 @@ async function runSubagent(config: SubagentRunConfig): Promise<void> {
 		appendControlEvent(event);
 		return true;
 	};
+	const updateStepModel = (flatIndex: number, model: string | undefined, thinking: string | undefined, now = Date.now()): void => {
+		const step = statusPayload.steps[flatIndex];
+		if (!step) return;
+		step.model = model;
+		step.thinking = thinking;
+		statusPayload.lastUpdate = now;
+		writeAtomicJson(statusPath, statusPayload);
+	};
 	const updateStepFromChildEvent = (flatIndex: number, event: ChildEvent): void => {
 		const step = statusPayload.steps[flatIndex];
 		if (!step) return;
@@ -1332,6 +1355,7 @@ async function runSubagent(config: SubagentRunConfig): Promise<void> {
 							registerInterrupt: (interrupt) => {
 								activeChildInterrupt = interrupt;
 							},
+							onAttemptStart: (attempt) => updateStepModel(fi, attempt.model, attempt.thinking),
 							onChildEvent: (event) => updateStepFromChildEvent(fi, event),
 						});
 						if (task.sessionFile) {
@@ -1346,6 +1370,7 @@ async function runSubagent(config: SubagentRunConfig): Promise<void> {
 						statusPayload.steps[fi].durationMs = taskDuration;
 						statusPayload.steps[fi].exitCode = singleResult.exitCode;
 						statusPayload.steps[fi].model = singleResult.model;
+						statusPayload.steps[fi].thinking = resolveEffectiveThinking(singleResult.model, statusPayload.steps[fi].thinking);
 						statusPayload.steps[fi].attemptedModels = singleResult.attemptedModels;
 						statusPayload.steps[fi].modelAttempts = singleResult.modelAttempts;
 						statusPayload.steps[fi].error = singleResult.error;
@@ -1475,6 +1500,7 @@ async function runSubagent(config: SubagentRunConfig): Promise<void> {
 				registerInterrupt: (interrupt) => {
 					activeChildInterrupt = interrupt;
 				},
+				onAttemptStart: (attempt) => updateStepModel(flatIndex, attempt.model, attempt.thinking),
 				onChildEvent: (event) => updateStepFromChildEvent(flatIndex, event),
 			});
 			if (seqStep.sessionFile) {
@@ -1522,6 +1548,7 @@ async function runSubagent(config: SubagentRunConfig): Promise<void> {
 			statusPayload.steps[flatIndex].durationMs = stepEndTime - stepStartTime;
 			statusPayload.steps[flatIndex].exitCode = singleResult.exitCode;
 			statusPayload.steps[flatIndex].model = singleResult.model;
+			statusPayload.steps[flatIndex].thinking = resolveEffectiveThinking(singleResult.model, statusPayload.steps[flatIndex].thinking);
 			statusPayload.steps[flatIndex].attemptedModels = singleResult.attemptedModels;
 			statusPayload.steps[flatIndex].modelAttempts = singleResult.modelAttempts;
 			statusPayload.steps[flatIndex].error = singleResult.error;

package/src/runs/foreground/execution.ts

@@ -151,6 +151,7 @@ async function runSingleAttempt(
 		extensions: agent.extensions,
 		systemPrompt: shared.systemPrompt,
 		mcpDirectTools: agent.mcpDirectTools,
+		cwd: options.cwd ?? runtimeCwd,
 		promptFileStem: agent.name,
 		intercomSessionName: options.intercomSessionName,
 		orchestratorIntercomTarget: options.orchestratorIntercomTarget,
@@ -207,6 +208,7 @@ async function runSingleAttempt(
 			cwd: options.cwd ?? runtimeCwd,
 			env: spawnEnv,
 			stdio: ["ignore", "pipe", "pipe"],
+			windowsHide: true,
 		});
 		const jsonlWriter = createJsonlWriter(shared.jsonlPath, proc.stdout);
 		let buf = "";
@@ -214,6 +216,7 @@ async function runSingleAttempt(
 		let settled = false;
 		let detached = false;
 		let intercomStarted = false;
+		let assistantError: string | undefined;
 		let removeAbortListener: (() => void) | undefined;
 		let removeInterruptListener: (() => void) | undefined;
 		let activityTimer: NodeJS.Timeout | undefined;
@@ -259,7 +262,7 @@ async function runSingleAttempt(
 				const termSent = trySignalChild(proc, "SIGTERM");
 				if (!termSent) return;
 				forcedTerminationSignal = true;
-				if (!cleanTerminalAssistantStopReceived) {
+				if (!cleanTerminalAssistantStopReceived && !assistantError) {
 					result.error = result.error ?? `Subagent process did not exit within ${FINAL_STOP_GRACE_MS}ms after its final message. Forcing termination.`;
 				}
 				finalHardKillTimer = setTimeout(() => {
@@ -465,13 +468,15 @@ async function runSingleAttempt(
 						progress.tokens = result.usage.input + result.usage.output;
 					}
 					if (!result.model && evt.message.model) result.model = evt.message.model;
-					if (evt.message.errorMessage) result.error = evt.message.errorMessage;
-					appendRecentOutput(progress, extractTextFromContent(evt.message.content).split("\n").slice(-10));
+					if (evt.message.errorMessage) assistantError = evt.message.errorMessage;
+					const assistantText = extractTextFromContent(evt.message.content);
+					appendRecentOutput(progress, assistantText.split("\n").slice(-10));
 					// Final assistant message: start the exit drain window.
 					const stopReason = (evt.message as { stopReason?: string }).stopReason;
 					const hasToolCall = Array.isArray(evt.message.content)
 						&& evt.message.content.some((part) => (part as { type?: string }).type === "toolCall");
 					if (stopReason === "stop" && !hasToolCall) {
+						if (!evt.message.errorMessage && assistantText.trim()) assistantError = undefined;
 						cleanTerminalAssistantStopReceived ||= !evt.message.errorMessage;
 						startFinalDrain();
 					}
@@ -551,6 +556,7 @@ async function runSingleAttempt(
 			}
 			processClosed = true;
 			if (buf.trim()) processLine(buf);
+			if (!result.error && assistantError) result.error = assistantError;
 			const forcedDrainAfterFinalSuccess = forcedTerminationSignal && cleanTerminalAssistantStopReceived && !result.error;
 			if (code !== 0 && stderrBuf.trim() && !result.error && !forcedDrainAfterFinalSuccess) {
 				result.error = stderrBuf.trim();
@@ -662,8 +668,14 @@ async function runSingleAttempt(
 	};
 
 	let fullOutput = getFinalOutput(result.messages);
-	const completionGuard = result.exitCode === 0 && !result.error
-		? evaluateCompletionMutationGuard({ agent: agent.name, task, messages: result.messages })
+	const completionGuard = result.exitCode === 0 && !result.error && agent.completionGuard !== false
+		? evaluateCompletionMutationGuard({
+			agent: agent.name,
+			task,
+			messages: result.messages,
+			tools: agent.tools,
+			mcpDirectTools: agent.mcpDirectTools,
+		})
 		: undefined;
 	if (completionGuard?.triggered && !observedMutationAttempt) {
 		result.exitCode = 1;

package/src/runs/foreground/subagent-executor.ts

@@ -35,7 +35,7 @@ import { createForkContextResolver } from "../../shared/fork-context.ts";
 import { resolveCurrentSessionId } from "../../shared/session-identity.ts";
 import { applyIntercomBridgeToAgent, INTERCOM_BRIDGE_MARKER, resolveIntercomBridge, resolveIntercomSessionTarget, resolveSubagentIntercomTarget, type IntercomBridgeState } from "../../intercom/intercom-bridge.ts";
 import { formatControlIntercomMessage, formatControlNoticeMessage, resolveControlConfig, shouldNotifyControlEvent } from "../shared/subagent-control.ts";
-import { finalizeSingleOutput, injectSingleOutputInstruction, resolveSingleOutputPath, validateFileOnlyOutputMode } from "../shared/single-output.ts";
+import { finalizeSingleOutput, injectSingleOutputInstruction, normalizeSingleOutputOverride, resolveSingleOutputPath, validateFileOnlyOutputMode } from "../shared/single-output.ts";
 import { compactForegroundDetails, getSingleResultOutput, mapConcurrent, readStatus, resolveChildCwd } from "../../shared/utils.ts";
 import {
 	buildSubagentResultIntercomPayload,
@@ -979,7 +979,7 @@ function runAsyncPath(data: ExecutionContextData, deps: ExecutorDeps): AgentTool
 			};
 		}
 		const rawOutput = params.output !== undefined ? params.output : a.output;
-		const effectiveOutput: string | false | undefined = rawOutput === true ? a.output : (rawOutput as string | false | undefined);
+		const effectiveOutput = normalizeSingleOutputOverride(rawOutput, a.output);
 		const effectiveOutputMode = params.outputMode ?? "inline";
 		const normalizedSkills = normalizeSkillInput(params.skill);
 		const skills = normalizedSkills === false ? [] : normalizedSkills;
@@ -1716,7 +1716,7 @@ async function runSinglePath(data: ExecutionContextData, deps: ExecutorDeps): Pr
 	);
 	let skillOverride: string[] | false | undefined = normalizeSkillInput(params.skill);
 	const rawOutput = params.output !== undefined ? params.output : agentConfig.output;
-	let effectiveOutput: string | false | undefined = rawOutput === true ? agentConfig.output : (rawOutput as string | false | undefined);
+	let effectiveOutput = normalizeSingleOutputOverride(rawOutput, agentConfig.output);
 	const effectiveOutputMode = params.outputMode ?? "inline";
 	const currentMaxSubagentDepth = resolveCurrentMaxSubagentDepth(deps.config.maxSubagentDepth);
 	const maxSubagentDepth = resolveChildMaxSubagentDepth(currentMaxSubagentDepth, agentConfig.maxSubagentDepth);
@@ -1750,7 +1750,7 @@ async function runSinglePath(data: ExecutionContextData, deps: ExecutorDeps): Pr
 		task = result.templates[0]!;
 		const override = result.behaviorOverrides[0];
 		if (override?.model) modelOverride = override.model;
-		if (override?.output !== undefined) effectiveOutput = override.output;
+		if (override?.output !== undefined) effectiveOutput = normalizeSingleOutputOverride(override.output, agentConfig.output);
 		if (override?.skills !== undefined) skillOverride = override.skills;
 
 		if (result.runInBackground) {
@@ -2127,9 +2127,8 @@ export function createSubagentExecutor(deps: ExecutorDeps): {
 			return toExecutionErrorResult(effectiveParams, error);
 		}
 		const requestedAsync = effectiveParams.async ?? deps.asyncByDefault;
-		const backgroundRequestedWhileClarifying = hasTasks && requestedAsync && effectiveParams.clarify === true;
-		const effectiveAsync = requestedAsync
-			&& (hasChain ? effectiveParams.clarify === false : effectiveParams.clarify !== true);
+		const backgroundRequestedWhileClarifying = (hasChain || hasTasks) && requestedAsync && effectiveParams.clarify === true;
+		const effectiveAsync = requestedAsync && effectiveParams.clarify !== true;
 		const controlConfig = resolveControlConfig(deps.config.control, effectiveParams.control);
 
 		const artifactConfig: ArtifactConfig = {

package/src/runs/shared/completion-guard.ts

@@ -54,11 +54,24 @@ const GENERAL_IMPLEMENTATION_PATTERNS = [
 	/\b(?:update|add|remove|replace|delete|create)\s+(?:the\s+)?(?:file|files|code|source|implementation|test|tests|component|function|module|class|method|logic|import|imports|readme|docs?|changelog|package\.json|config|manifest|extension|prompt|command)\b/i,
 ];
 
+const READ_ONLY_BUILTIN_TOOLS = new Set([
+	"read",
+	"grep",
+	"find",
+	"ls",
+	"web_search",
+	"fetch_content",
+	"get_search_content",
+	"intercom",
+	"contact_supervisor",
+]);
 
 interface CompletionMutationGuardInput {
 	agent: string;
 	task: string;
 	messages: Message[];
+	tools?: string[];
+	mcpDirectTools?: string[];
 }
 
 interface CompletionMutationGuardResult {
@@ -83,6 +96,13 @@ function stripScopedNoEditConstraints(task: string): string {
 	return stripped;
 }
 
+function declaresOnlyReadOnlyTools(tools: string[] | undefined, mcpDirectTools: string[] | undefined): boolean {
+	return tools !== undefined
+		&& tools.length > 0
+		&& (mcpDirectTools?.length ?? 0) === 0
+		&& tools.every((tool) => READ_ONLY_BUILTIN_TOOLS.has(tool));
+}
+
 export function expectsImplementationMutation(agent: string, task: string): boolean {
 	const taskText = stripFrameworkInstructions(task);
 	const taskTextWithoutScopedConstraints = stripScopedNoEditConstraints(taskText);
@@ -115,7 +135,9 @@ export function hasMutationToolCall(messages: Message[]): boolean {
 }
 
 export function evaluateCompletionMutationGuard(input: CompletionMutationGuardInput): CompletionMutationGuardResult {
-	const expectedMutation = expectsImplementationMutation(input.agent, input.task);
+	const expectedMutation = declaresOnlyReadOnlyTools(input.tools, input.mcpDirectTools)
+		? false
+		: expectsImplementationMutation(input.agent, input.task);
 	const attemptedMutation = hasMutationToolCall(input.messages);
 	return {
 		expectedMutation,