ur-agent 1.80.6 → 1.80.8

package/CHANGELOG.md

@@ -1,5 +1,38 @@
 # Changelog
 
+## 1.80.8
+
+- Fixed provider-independent failed-first research delegation. Several model
+  families selected `general-purpose` even when their own worker brief said
+  read-only research, causing `TaskListRequired` before the research could
+  start. UR now safely downgrades that exact main-session contract to its
+  mechanically read-only shipped `Explore` agent before task gating.
+- Strengthened the shared Agent prompt and Explore description so models choose
+  `subagent_type="Explore"` directly for task-free research. The compatibility
+  downgrade also recognizes “research task only; do not write files” wording.
+- Kept the boundary fail-closed: implementation prompts, custom definitions,
+  nested workers, named/team agents, worktrees, cwd overrides, and every other
+  general-purpose call still require an actionable task. Permission and hook
+  rewrites are revalidated before execution. Approve All remains unchanged.
+
+## 1.80.7
+
+- Fixed Ollama runs stopping with `response returned unavailable tool
+  "WebSearch"`. Syntactically valid but unavailable native and text-form calls
+  now reach UR's guarded executor, which returns a recoverable result without
+  executing the tool and tells the model to use an available alternative or
+  return useful partial work. Identical retries are bounded; malformed names
+  and arguments still fail closed.
+- Extended task-free read-only research beyond Plan Mode. The main session may
+  launch UR's exact shipped `Explore` and `Plan` agents before tasks exist in
+  every permission mode, and those workers are forced into plan permissions
+  even when the parent uses Accept Edits or Approve All. Custom, write-capable,
+  nested, team, and worktree agents still require an actionable parent task.
+- Removed the global deprecated-alias fallback from tool execution. Aliases
+  continue to work for tools present in the active profile, but can no longer
+  revive a tool deliberately omitted from a worker. Approve All itself remains
+  supported and unchanged.
+
 ## 1.80.6
 
 - Fixed Plan Mode's failed-first research delegation. UR's shipped read-only

package/dist/cli.js

@@ -88903,6 +88903,9 @@ function parseTextToolCalls(text, options = {}) {
       return call;
     const name = reconcileToolName(call.name, options.availableToolNames);
     if (!hasTool(options.availableToolNames, name)) {
+      if (options.preserveUnavailableToolCalls) {
+        return name === call.name ? call : { ...call, name };
+      }
       throw new KimiToolCallParseError(`Kimi returned unavailable tool "${name}"`);
     }
     return name === call.name ? call : { ...call, name };
@@ -90333,7 +90336,8 @@ function ollamaResponseToURHQMessage(response, params, textToolFallbackAllowed)
   const rawText = response.message?.content ?? "";
   const parsedText = textToolFallbackAllowed ? parseTextToolCalls(rawText, {
     availableToolNames,
-    parseBareJsonToolCalls: true
+    parseBareJsonToolCalls: true,
+    preserveUnavailableToolCalls: true
   }) : { text: rawText, toolCalls: [] };
   const text = parsedText.text;
   const textToolCalls = [...parsedText.toolCalls];
@@ -90423,7 +90427,7 @@ function normalizeOllamaToolUses(structured, textCalls, availableToolNames, cont
     }
     const name = reconcileToolName(rawName, availableToolNames);
     if (!availableToolNames.has(name)) {
-      throw new ProviderResponseParseError(`${context} returned unavailable tool "${name}"`, { rawName, availableToolNames: [...availableToolNames] });
+      logForDebugging(`${context} preserved unavailable tool "${name}" for guarded rejection`, { level: "warn" });
     }
     const key = `${name}\x00${toolArgsKey(input)}`;
     if (seen.has(key)) {
@@ -107784,7 +107788,7 @@ var init_auth = __esm(() => {
 
 // src/utils/userAgent.ts
 function getURCodeUserAgent() {
-  return `ur/${"1.80.6"}`;
+  return `ur/${"1.80.8"}`;
 }
 
 // src/utils/workloadContext.ts
@@ -107806,7 +107810,7 @@ function getUserAgent() {
   const clientApp = process.env.UR_AGENT_SDK_CLIENT_APP ? `, client-app/${process.env.UR_AGENT_SDK_CLIENT_APP}` : "";
   const workload = getWorkload();
   const workloadSuffix = workload ? `, workload/${workload}` : "";
-  return `ur-cli/${"1.80.6"} (${process.env.USER_TYPE}, ${process.env.UR_CODE_ENTRYPOINT ?? "cli"}${agentSdkVersion}${clientApp}${workloadSuffix})`;
+  return `ur-cli/${"1.80.8"} (${process.env.USER_TYPE}, ${process.env.UR_CODE_ENTRYPOINT ?? "cli"}${agentSdkVersion}${clientApp}${workloadSuffix})`;
 }
 function getMCPUserAgent() {
   const parts = [];
@@ -107820,7 +107824,7 @@ function getMCPUserAgent() {
     parts.push(`client-app/${process.env.UR_AGENT_SDK_CLIENT_APP}`);
   }
   const suffix = parts.length > 0 ? ` (${parts.join(", ")})` : "";
-  return `ur/${"1.80.6"}${suffix}`;
+  return `ur/${"1.80.8"}${suffix}`;
 }
 function getWebFetchUserAgent() {
   return `UR-User (${getURCodeUserAgent()})`;
@@ -107958,7 +107962,7 @@ var init_user = __esm(() => {
       deviceId,
       sessionId: getSessionId(),
       email: getEmail(),
-      appVersion: "1.80.6",
+      appVersion: "1.80.8",
       platform: getHostPlatformForAnalytics(),
       organizationUuid,
       accountUuid,
@@ -115845,7 +115849,7 @@ var init_metadata = __esm(() => {
   COMPOUND_OPERATOR_REGEX = /\s*(?:&&|\|\||[;|])\s*/;
   WHITESPACE_REGEX = /\s+/;
   getVersionBase = memoize_default(() => {
-    const match = "1.80.6".match(/^\d+\.\d+\.\d+(?:-[a-z]+)?/);
+    const match = "1.80.8".match(/^\d+\.\d+\.\d+(?:-[a-z]+)?/);
     return match ? match[0] : undefined;
   });
   buildEnvContext = memoize_default(async () => {
@@ -115885,7 +115889,7 @@ var init_metadata = __esm(() => {
       isGithubAction: isEnvTruthy(process.env.GITHUB_ACTIONS),
       isURCodeAction: isEnvTruthy(process.env.UR_CODE_ACTION),
       isURAiAuth: isURAISubscriber(),
-      version: "1.80.6",
+      version: "1.80.8",
       versionBase: getVersionBase(),
       buildTime: "",
       deploymentEnvironment: env2.detectDeploymentEnvironment(),
@@ -116555,7 +116559,7 @@ function initialize1PEventLogging() {
   const platform2 = getPlatform();
   const attributes = {
     [import_semantic_conventions4.ATTR_SERVICE_NAME]: "ur",
-    [import_semantic_conventions4.ATTR_SERVICE_VERSION]: "1.80.6"
+    [import_semantic_conventions4.ATTR_SERVICE_VERSION]: "1.80.8"
   };
   if (platform2 === "wsl") {
     const wslVersion = getWslVersion();
@@ -116583,7 +116587,7 @@ function initialize1PEventLogging() {
       })
     ]
   });
-  firstPartyEventLogger = firstPartyEventLoggerProvider.getLogger("com.urhq.ur.events", "1.80.6");
+  firstPartyEventLogger = firstPartyEventLoggerProvider.getLogger("com.urhq.ur.events", "1.80.8");
 }
 async function reinitialize1PEventLoggingIfConfigChanged() {
   if (!is1PEventLoggingEnabled() || !firstPartyEventLoggerProvider) {
@@ -126546,7 +126550,7 @@ function formatA2AAgentCard(options = {}, pretty = true) {
 function formatA2AV1AgentCard(options = {}, pretty = true) {
   return JSON.stringify(buildA2AV1AgentCard(options), null, pretty ? 2 : 0);
 }
-var urVersion = "1.80.6", researchSnapshotDate = "2026-08-10", coverage, priorityRoadmap;
+var urVersion = "1.80.8", researchSnapshotDate = "2026-08-10", coverage, priorityRoadmap;
 var init_trends = __esm(() => {
   init_a2aCardSignature();
   coverage = [
@@ -129436,7 +129440,7 @@ function getAttributionHeader(fingerprint) {
   if (!isAttributionHeaderEnabled()) {
     return "";
   }
-  const version2 = `${"1.80.6"}.${fingerprint}`;
+  const version2 = `${"1.80.8"}.${fingerprint}`;
   const entrypoint = process.env.UR_CODE_ENTRYPOINT ?? "unknown";
   const cch = "";
   const workload = getWorkload();
@@ -174266,7 +174270,7 @@ NOTE: You are meant to be a fast agent that returns output as quickly as possibl
 
 Complete the user's search request efficiently and report your findings clearly.`;
 }
-var EXPLORE_AGENT_MIN_QUERIES = 3, EXPLORE_WHEN_TO_USE = 'Fast agent specialized for exploring codebases. Use this when you need to quickly find files by patterns (eg. "src/components/**/*.tsx"), search code for keywords (eg. "API endpoints"), or answer questions about the codebase (eg. "how do API endpoints work?"). When calling this agent, specify the desired thoroughness level: "quick" for basic searches, "medium" for moderate exploration, or "very thorough" for comprehensive analysis across multiple locations and naming conventions.', EXPLORE_AGENT;
+var EXPLORE_AGENT_MIN_QUERIES = 3, EXPLORE_WHEN_TO_USE = 'Read-only research and exploration agent. Use this for external research, source gathering, audits, codebase exploration, file-pattern searches (eg. "src/components/**/*.tsx"), code searches (eg. "API endpoints"), and questions about how a repository works. Prefer this over general-purpose whenever the worker must not modify files, especially before an actionable task exists. Specify the desired thoroughness: "quick", "medium", or "very thorough".', EXPLORE_AGENT;
 var init_exploreAgent = __esm(() => {
   init_prompt3();
   init_prompt4();
@@ -184632,7 +184636,7 @@ var init_projectSafety = __esm(() => {
 function getInstruments() {
   if (instruments)
     return instruments;
-  const meter = import_api10.metrics.getMeter("ur-agent.gen_ai", "1.80.6");
+  const meter = import_api10.metrics.getMeter("ur-agent.gen_ai", "1.80.8");
   instruments = {
     operationDuration: meter.createHistogram("gen_ai.client.operation.duration", {
       description: "GenAI operation duration.",
@@ -184730,7 +184734,7 @@ function genAiAgentAttributes() {
     "gen_ai.operation.name": GEN_AI_OPERATION_INVOKE_AGENT,
     "gen_ai.provider.name": "ur",
     "gen_ai.agent.name": "UR-Nexus",
-    "gen_ai.agent.version": "1.80.6"
+    "gen_ai.agent.version": "1.80.8"
   };
 }
 function genAiWorkflowAttributes(workflowName, workflowRunId) {
@@ -184751,7 +184755,7 @@ function genAiWorkflowAttributes(workflowName, workflowRunId) {
 function startGenAiWorkflowSpan(workflowName, workflowRunId) {
   const attributes = genAiWorkflowAttributes(workflowName, workflowRunId);
   const name = typeof attributes["gen_ai.workflow.name"] === "string" ? `invoke_workflow ${attributes["gen_ai.workflow.name"]}` : GEN_AI_OPERATION_INVOKE_WORKFLOW;
-  return import_api10.trace.getTracer("ur-agent.gen_ai", "1.80.6").startSpan(name, { kind: import_api10.SpanKind.INTERNAL, attributes });
+  return import_api10.trace.getTracer("ur-agent.gen_ai", "1.80.8").startSpan(name, { kind: import_api10.SpanKind.INTERNAL, attributes });
 }
 function endGenAiWorkflowSpan(span, options2 = {}) {
   try {
@@ -184789,7 +184793,7 @@ function startGenAiMemorySpan(operation, options2 = {}) {
   if (options2.recordCount !== undefined && Number.isInteger(options2.recordCount) && options2.recordCount >= 0) {
     attributes["gen_ai.memory.record.count"] = options2.recordCount;
   }
-  return import_api10.trace.getTracer("ur-agent.gen_ai", "1.80.6").startSpan(operation, { kind: import_api10.SpanKind.INTERNAL, attributes });
+  return import_api10.trace.getTracer("ur-agent.gen_ai", "1.80.8").startSpan(operation, { kind: import_api10.SpanKind.INTERNAL, attributes });
 }
 function endGenAiMemorySpan(span, options2 = {}) {
   try {
@@ -278504,7 +278508,7 @@ function getTelemetryAttributes() {
     attributes["session.id"] = sessionId;
   }
   if (shouldIncludeAttribute("OTEL_METRICS_INCLUDE_VERSION")) {
-    attributes["app.version"] = "1.80.6";
+    attributes["app.version"] = "1.80.8";
   }
   const oauthAccount = getOauthAccountInfo();
   if (oauthAccount) {
@@ -319495,7 +319499,7 @@ function getInstallationEnv() {
   return;
 }
 function getURCodeVersion() {
-  return "1.80.6";
+  return "1.80.8";
 }
 async function getInstalledVSCodeExtensionVersion(command) {
   const { stdout } = await execFileNoThrow(command, ["--list-extensions", "--show-versions"], {
@@ -326865,7 +326869,7 @@ async function setupSdkMcpClients(sdkMcpConfigs, sendMcpMessage) {
     const client2 = new Client({
       name: "ur",
       title: "UR",
-      version: "1.80.6",
+      version: "1.80.8",
       description: "UR-Nexus autonomous engineering workflow engine",
       websiteUrl: PRODUCT_URL
     }, {
@@ -327226,7 +327230,7 @@ var init_client5 = __esm(() => {
       const client2 = new Client({
         name: "ur",
         title: "UR",
-        version: "1.80.6",
+        version: "1.80.8",
         description: "UR-Nexus autonomous engineering workflow engine",
         websiteUrl: PRODUCT_URL
       }, {
@@ -329242,6 +329246,36 @@ var init_agentToolUtils = __esm(() => {
   }));
 });
 
+// src/tools/AgentTool/readOnlyAgents.ts
+function isShippedReadOnlyAgentDefinition(agent) {
+  return SHIPPED_READ_ONLY_AGENT_TYPES.has(agent.agentType) && agent.source === "built-in" && agent.permissionMode === "plan";
+}
+function normalizeReadOnlyResearchDelegation(input, activeAgents, isNestedAgent) {
+  if (isNestedAgent || input.subagent_type !== "general-purpose" || input.name !== undefined || input.team_name !== undefined || input.isolation !== undefined || input.cwd !== undefined || typeof input.prompt !== "string" || !READ_ONLY_RESEARCH_MARKER.test(input.prompt) || !RESEARCH_INTENT_MARKER.test(input.prompt)) {
+    return input;
+  }
+  const exploreAgent = activeAgents.find((agent) => agent.agentType === "Explore" && isShippedReadOnlyAgentDefinition(agent));
+  if (!exploreAgent)
+    return input;
+  return { ...input, subagent_type: exploreAgent.agentType };
+}
+function shouldApplyAgentDefinitionPermissionMode(agent, parentMode, transcriptClassifierEnabled) {
+  if (!agent.permissionMode)
+    return false;
+  if (isShippedReadOnlyAgentDefinition(agent))
+    return true;
+  return parentMode !== "bypassPermissions" && parentMode !== "acceptEdits" && !(transcriptClassifierEnabled && parentMode === "auto");
+}
+var SHIPPED_READ_ONLY_AGENT_TYPES, READ_ONLY_RESEARCH_MARKER, RESEARCH_INTENT_MARKER;
+var init_readOnlyAgents = __esm(() => {
+  SHIPPED_READ_ONLY_AGENT_TYPES = new Set([
+    "Explore",
+    "Plan"
+  ]);
+  READ_ONLY_RESEARCH_MARKER = /\bread[\s-]?only\b|\b(?:do not|don't|must not|never)\s+(?:modify|write|edit|create|change)\s+(?:any\s+)?(?:code\s+)?files?\b|\bno\s+(?:file|workspace)\s+(?:changes|modifications)\b/i;
+  RESEARCH_INTENT_MARKER = /\b(?:research|investigat\w*|explor\w*|analy[sz]\w*|audit\w*)\b/i;
+});
+
 // src/components/AgentProgressLine.tsx
 function getAgentProgressStatus({
   isResolved,
@@ -339964,7 +339998,7 @@ async function createRuntime() {
   bootstrapTelemetry();
   const resource = defaultResource().merge(detectResources({ detectors: [envDetector] })).merge(resourceFromAttributes({
     [import_semantic_conventions7.ATTR_SERVICE_NAME]: "ur-agent",
-    [import_semantic_conventions7.ATTR_SERVICE_VERSION]: "1.80.6"
+    [import_semantic_conventions7.ATTR_SERVICE_VERSION]: "1.80.8"
   }));
   const tracerProvider = exporters.traces.length > 0 ? new BasicTracerProvider({
     resource,
@@ -339997,11 +340031,11 @@ async function createRuntime() {
   setMeterProvider(meterProvider);
   setLoggerProvider(loggerProvider);
   if (meterProvider) {
-    const meter = meterProvider.getMeter("ur-agent", "1.80.6");
+    const meter = meterProvider.getMeter("ur-agent", "1.80.8");
     setMeter(meter, (name, options2) => meter.createCounter(name, options2));
   }
   if (loggerProvider) {
-    setEventLogger(loggerProvider.getLogger("ur-agent.events", "1.80.6"));
+    setEventLogger(loggerProvider.getLogger("ur-agent.events", "1.80.8"));
   }
   if (!cleanupRegistered3) {
     cleanupRegistered3 = true;
@@ -340663,9 +340697,9 @@ async function assertMinVersion() {
   if (false) {}
   try {
     const versionConfig = await getDynamicConfig_BLOCKS_ON_INIT("tengu_version_config", { minVersion: "0.0.0" });
-    if (versionConfig.minVersion && lt("1.80.6", versionConfig.minVersion)) {
+    if (versionConfig.minVersion && lt("1.80.8", versionConfig.minVersion)) {
       console.error(`
-It looks like your version of UR (${"1.80.6"}) needs an update.
+It looks like your version of UR (${"1.80.8"}) needs an update.
 A newer version (${versionConfig.minVersion} or higher) is required to continue.
 
 To update, please run:
@@ -340881,7 +340915,7 @@ async function installGlobalPackage(specificVersion) {
     logError2(new AutoUpdaterError("Another process is currently installing an update"));
     logEvent("tengu_auto_updater_lock_contention", {
       pid: process.pid,
-      currentVersion: "1.80.6"
+      currentVersion: "1.80.8"
     });
     return "in_progress";
   }
@@ -340890,7 +340924,7 @@ async function installGlobalPackage(specificVersion) {
     if (!env2.isRunningWithBun() && env2.isNpmFromWindowsPath()) {
       logError2(new Error("Windows NPM detected in WSL environment"));
       logEvent("tengu_auto_updater_windows_npm_in_wsl", {
-        currentVersion: "1.80.6"
+        currentVersion: "1.80.8"
       });
       console.error(`
 Error: Windows NPM detected in WSL
@@ -341425,7 +341459,7 @@ function detectLinuxGlobPatternWarnings() {
 }
 async function getDoctorDiagnostic() {
   const installationType = await getCurrentInstallationType();
-  const version2 = typeof MACRO !== "undefined" ? "1.80.6" : "unknown";
+  const version2 = typeof MACRO !== "undefined" ? "1.80.8" : "unknown";
   const installationPath = await getInstallationPath();
   const invokedBinary = getInvokedBinary();
   const multipleInstallations = await detectMultipleInstallations();
@@ -342360,8 +342394,8 @@ async function updateLatest(channelOrVersion, forceReinstall = false) {
     const maxVersion = await getMaxVersion();
     if (maxVersion && gt(version2, maxVersion)) {
       logForDebugging(`Native installer: maxVersion ${maxVersion} is set, capping update from ${version2} to ${maxVersion}`);
-      if (gte("1.80.6", maxVersion)) {
-        logForDebugging(`Native installer: current version ${"1.80.6"} is already at or above maxVersion ${maxVersion}, skipping update`);
+      if (gte("1.80.8", maxVersion)) {
+        logForDebugging(`Native installer: current version ${"1.80.8"} is already at or above maxVersion ${maxVersion}, skipping update`);
         logEvent("tengu_native_update_skipped_max_version", {
           latency_ms: Date.now() - startTime,
           max_version: maxVersion,
@@ -342372,7 +342406,7 @@ async function updateLatest(channelOrVersion, forceReinstall = false) {
       version2 = maxVersion;
     }
   }
-  if (!forceReinstall && version2 === "1.80.6" && await versionIsAvailable(version2) && await isPossibleURBinary(executablePath)) {
+  if (!forceReinstall && version2 === "1.80.8" && await versionIsAvailable(version2) && await isPossibleURBinary(executablePath)) {
     logForDebugging(`Found ${version2} at ${executablePath}, skipping install`);
     logEvent("tengu_native_update_complete", {
       latency_ms: Date.now() - startTime,
@@ -363462,7 +363496,7 @@ async function* runAgent({
   const agentGetAppState = () => {
     const state = toolUseContext.getAppState();
     let toolPermissionContext = state.toolPermissionContext;
-    if (agentPermissionMode && state.toolPermissionContext.mode !== "bypassPermissions" && state.toolPermissionContext.mode !== "acceptEdits" && true) {
+    if (agentPermissionMode && shouldApplyAgentDefinitionPermissionMode(agentDefinition, state.toolPermissionContext.mode, false)) {
       toolPermissionContext = {
         ...toolPermissionContext,
         mode: agentPermissionMode
@@ -363755,6 +363789,7 @@ var init_runAgent = __esm(() => {
   init_uuid();
   init_agentToolUtils();
   init_loadAgentsDir();
+  init_readOnlyAgents();
 });
 
 // src/services/AgentSummary/agentSummary.ts
@@ -396228,7 +396263,7 @@ function checkTaskListGate(input) {
     return { allowed: true };
   if (input.isPlanningArtifact === true)
     return { allowed: true };
-  if (input.isReadOnlyPlanningDelegation === true)
+  if (input.isReadOnlyBuiltInDelegation === true)
     return { allowed: true };
   const isMutating = input.isMutating ?? isMutatingTool2(input.toolName);
   if (!isMutating)
@@ -406681,7 +406716,7 @@ async function getPrompt9(agentDefinitions, isCoordinator, allowedAgentTypes) {
 ## When to fork
 
 Fork yourself (omit \`subagent_type\`) when the intermediate tool output isn't worth keeping in your context. The criterion is qualitative \u2014 "will I need this output again" \u2014 not task size.
-- **Research**: fork open-ended questions. If research can be broken into independent questions, launch parallel forks in one message. A fork beats a fresh subagent for this \u2014 it inherits context and shares your cache.
+- **Research after tasks exist**: fork open-ended questions when inherited context is valuable. Before an actionable task exists, use the shipped \`Explore\` agent instead; a fork or general-purpose agent is write-capable and will be task-gated. If research can be broken into independent questions, launch parallel \`Explore\` agents in one message.
 - **Implementation**: fork only a bounded branch with explicit file scope. Parallel implementation forks must use isolated worktrees based on the exact clean starting revision; otherwise keep shared-checkout writers sequential. Do research before jumping to implementation.
 
 Forks are cheap because they share your prompt cache. Don't set \`model\` on a fork \u2014 a different model can't reuse the parent's cache. Pass a short \`name\` (one or two words, lowercase) so the user can see the fork in the teams panel and steer it mid-run.
@@ -413027,7 +413062,7 @@ function isAnyTracingEnabled() {
   return isTelemetryEnabled() || isEnhancedTelemetryEnabled() || isBetaTracingEnabled();
 }
 function getTracer() {
-  return import_api39.trace.getTracer("ur-agent.gen_ai", "1.80.6");
+  return import_api39.trace.getTracer("ur-agent.gen_ai", "1.80.8");
 }
 function createSpanAttributes(spanType, customAttributes = {}) {
   const baseAttributes = getTelemetryAttributes();
@@ -414482,15 +414517,15 @@ function isCurrentPlanFileMutation(toolName, input, context5) {
     return false;
   }
 }
-function isReadOnlyPlanningDelegation(toolName, input, context5) {
-  if (context5.agentId || context5.getAppState().toolPermissionContext.mode !== "plan" || toolName !== AGENT_TOOL_NAME && toolName !== LEGACY_AGENT_TOOL_NAME || !input || typeof input !== "object" || Array.isArray(input)) {
+function isReadOnlyBuiltInDelegation(toolName, input, context5) {
+  if (context5.agentId || toolName !== AGENT_TOOL_NAME && toolName !== LEGACY_AGENT_TOOL_NAME || !input || typeof input !== "object" || Array.isArray(input)) {
     return false;
   }
   const delegation = input;
-  if (typeof delegation.subagent_type !== "string" || !READ_ONLY_PLANNING_AGENT_TYPES.has(delegation.subagent_type) || delegation.name !== undefined || delegation.team_name !== undefined || delegation.isolation !== undefined) {
+  if (typeof delegation.subagent_type !== "string" || delegation.name !== undefined || delegation.team_name !== undefined || delegation.isolation !== undefined) {
     return false;
   }
-  return context5.options.agentDefinitions.activeAgents.some((agent) => agent.agentType === delegation.subagent_type && agent.source === "built-in" && agent.permissionMode === "plan");
+  return context5.options.agentDefinitions.activeAgents.some((agent) => agent.agentType === delegation.subagent_type && isShippedReadOnlyAgentDefinition(agent));
 }
 function getStopHookInfo(attachment) {
   if (typeof attachment !== "object" || attachment === null || !("command" in attachment) || typeof attachment.command !== "string" || !("durationMs" in attachment) || typeof attachment.durationMs !== "number") {
@@ -414604,13 +414639,7 @@ function getMcpServerBaseUrlFromToolName(toolName, mcpClients) {
 }
 async function* runToolUse(toolUse, assistantMessage, canUseTool, toolUseContext) {
   const toolName = toolUse.name;
-  let tool = findToolByName(toolUseContext.options.tools, toolName);
-  if (!tool) {
-    const fallbackTool = findToolByName(getAllBaseTools(), toolName);
-    if (fallbackTool && fallbackTool.aliases?.includes(toolName)) {
-      tool = fallbackTool;
-    }
-  }
+  const tool = findToolByName(toolUseContext.options.tools, toolName);
   const messageId = assistantMessage.message?.id;
   if (typeof messageId !== "string" || messageId.length === 0) {
     throw new Error(`Cannot execute tool_use ${toolUse.id}: assistant message has no id`);
@@ -414620,7 +414649,7 @@ async function* runToolUse(toolUse, assistantMessage, canUseTool, toolUseContext
   const mcpServerBaseUrl = getMcpServerBaseUrlFromToolName(toolName, toolUseContext.options.mcpClients);
   if (!tool) {
     const callSig = callSignature(toolName, toolUse.input, repeatedFailureScope(toolUseContext, messageId));
-    const repeat2 = checkRepeatedFailure(callSig);
+    const repeat2 = checkRepeatedFailure(callSig, UNKNOWN_TOOL_REPEAT_POLICY);
     if (repeat2.action === "abort") {
       throw new RepeatedToolFailureAbort(`Repeated tool failure: ${repeat2.reason}`);
     }
@@ -414662,17 +414691,18 @@ async function* runToolUse(toolUse, assistantMessage, canUseTool, toolUseContext
       },
       ...mcpToolDetailsForAnalytics(toolName, mcpServerType, mcpServerBaseUrl)
     });
+    const unavailableMessage = `Tool "${toolName}" is not available in this agent's active tool profile. Do not retry this tool unchanged. Continue with an available tool, or return the useful partial result so the parent agent can proceed.`;
     yield {
       message: createUserMessage({
         content: [
           {
             type: "tool_result",
-            content: `<tool_use_error>Error: No such tool available: ${toolName}</tool_use_error>`,
+            content: `<tool_use_error>UnavailableTool: ${unavailableMessage}</tool_use_error>`,
             is_error: true,
             tool_use_id: toolUse.id
           }
         ],
-        toolUseResult: `Error: No such tool available: ${toolName}`,
+        toolUseResult: `UnavailableTool: ${unavailableMessage}`,
         sourceToolAssistantUUID: assistantMessage.uuid
       })
     };
@@ -414824,6 +414854,16 @@ async function checkPermissionsAndCallTool(tool, toolUseID, input, toolUseContex
       }
     }
   }
+  if (parsedInput.success && (tool.name === AGENT_TOOL_NAME || tool.name === LEGACY_AGENT_TOOL_NAME)) {
+    const normalizedDelegation = normalizeReadOnlyResearchDelegation(parsedInput.data, toolUseContext.options.agentDefinitions.activeAgents, Boolean(toolUseContext.agentId));
+    if (normalizedDelegation !== parsedInput.data) {
+      input = normalizedDelegation;
+      parsedInput = tool.inputSchema.safeParse(input);
+      logEvent("tengu_agent_read_only_research_normalized", {
+        toolName: sanitizeToolNameForAnalytics(tool.name)
+      });
+    }
+  }
   let callSig = callSignature(tool.name, input, repeatedFailureScope(toolUseContext, messageId));
   const repeat2 = checkRepeatedFailure(callSig);
   if (repeat2.action !== "allow") {
@@ -414957,7 +414997,7 @@ async function checkPermissionsAndCallTool(tool, toolUseID, input, toolUseContex
     requiresTaskList: taskListRun?.requiresTaskList,
     requirementReason: taskListRun?.requirementReason,
     isPlanningArtifact: isCurrentPlanFileMutation(tool.name, parsedInput.data, toolUseContext),
-    isReadOnlyPlanningDelegation: isReadOnlyPlanningDelegation(tool.name, parsedInput.data, toolUseContext),
+    isReadOnlyBuiltInDelegation: isReadOnlyBuiltInDelegation(tool.name, parsedInput.data, toolUseContext),
     taskListWriterAvailable: toolUseContext.options.tools.some((candidate) => candidate.name === "TaskCreate")
   });
   if (gate.allowed === false) {
@@ -415319,7 +415359,7 @@ async function checkPermissionsAndCallTool(tool, toolUseID, input, toolUseContex
       requiresTaskList: taskListRun?.requiresTaskList,
       requirementReason: taskListRun?.requirementReason,
       isPlanningArtifact: isCurrentPlanFileMutation(tool.name, finalParsedInput.data, toolUseContext),
-      isReadOnlyPlanningDelegation: isReadOnlyPlanningDelegation(tool.name, finalParsedInput.data, toolUseContext),
+      isReadOnlyBuiltInDelegation: isReadOnlyBuiltInDelegation(tool.name, finalParsedInput.data, toolUseContext),
       taskListWriterAvailable: toolUseContext.options.tools.some((candidate) => candidate.name === "TaskCreate")
     });
     if (finalGate.allowed === false) {
@@ -415729,7 +415769,7 @@ async function checkPermissionsAndCallTool(tool, toolUseID, input, toolUseContex
     }
   }
 }
-var READ_ONLY_PLANNING_AGENT_TYPES, HOOK_TIMING_DISPLAY_THRESHOLD_MS2 = 500, SLOW_PHASE_LOG_THRESHOLD_MS = 2000;
+var UNKNOWN_TOOL_REPEAT_POLICY, HOOK_TIMING_DISPLAY_THRESHOLD_MS2 = 500, SLOW_PHASE_LOG_THRESHOLD_MS = 2000;
 var init_toolExecution = __esm(() => {
   init_analytics();
   init_metadata();
@@ -415741,11 +415781,11 @@ var init_toolExecution = __esm(() => {
   init_prompt();
   init_bashPermissions();
   init_constants2();
+  init_readOnlyAgents();
   init_prompt3();
   init_prompt4();
   init_gitOperationTracking();
   init_prompt8();
-  init_tools2();
   init_attachments2();
   init_debug();
   init_errors();
@@ -415771,7 +415811,11 @@ var init_toolExecution = __esm(() => {
   init_mcpStringUtils();
   init_utils3();
   init_toolHooks();
-  READ_ONLY_PLANNING_AGENT_TYPES = new Set(["Explore", "Plan"]);
+  UNKNOWN_TOOL_REPEAT_POLICY = {
+    enabled: true,
+    limit: 1,
+    abortAfter: 3
+  };
 });
 
 // src/services/tools/StreamingToolExecutor.ts
@@ -443147,7 +443191,7 @@ function Feedback({
       platform: env2.platform,
       gitRepo: envInfo.isGit,
       terminal: env2.terminal,
-      version: "1.80.6",
+      version: "1.80.8",
       transcript: normalizeMessagesForAPI(messages),
       errors: sanitizedErrors,
       lastApiRequest: getLastAPIRequest(),
@@ -443339,7 +443383,7 @@ function Feedback({
                       ", ",
                       env2.terminal,
                       ", v",
-                      "1.80.6"
+                      "1.80.8"
                     ]
                   }, undefined, true, undefined, this)
                 ]
@@ -443445,7 +443489,7 @@ ${sanitizedDescription}
 ` + `**Environment Info**
 ` + `- Platform: ${env2.platform}
 ` + `- Terminal: ${env2.terminal}
-` + `- Version: ${"1.80.6"}
+` + `- Version: ${"1.80.8"}
 ` + `- Feedback ID: ${feedbackId}
 ` + `
 **Errors**
@@ -446555,7 +446599,7 @@ function buildPrimarySection() {
   }, undefined, false, undefined, this);
   return [{
     label: "Version",
-    value: "1.80.6"
+    value: "1.80.8"
   }, {
     label: "Session name",
     value: nameValue
@@ -449937,7 +449981,7 @@ function Config({
         }
       }, undefined, false, undefined, this)
     }, undefined, false, undefined, this) : showSubmenu === "ChannelDowngrade" ? /* @__PURE__ */ jsx_dev_runtime179.jsxDEV(ChannelDowngradeDialog, {
-      currentVersion: "1.80.6",
+      currentVersion: "1.80.8",
       onChoice: (choice) => {
         setShowSubmenu(null);
         setTabsHidden(false);
@@ -449949,7 +449993,7 @@ function Config({
           autoUpdatesChannel: "stable"
         };
         if (choice === "stay") {
-          newSettings.minimumVersion = "1.80.6";
+          newSettings.minimumVersion = "1.80.8";
         }
         updateSettingsForSource("userSettings", newSettings);
         setSettingsData((prev_27) => ({
@@ -458258,7 +458302,7 @@ function HelpV2(t0) {
   let t6;
   if ($2[31] !== tabs) {
     t6 = /* @__PURE__ */ jsx_dev_runtime206.jsxDEV(Tabs, {
-      title: `UR v${"1.80.6"}`,
+      title: `UR v${"1.80.8"}`,
       color: "professionalBlue",
       defaultTab: "general",
       children: tabs
@@ -459191,7 +459235,7 @@ function buildToolUseContext(tools, readFileStateCache, toolPermissionContext, a
 async function handleInitialize(options2) {
   return {
     name: "UR",
-    version: "1.80.6",
+    version: "1.80.8",
     protocolVersion: "0.1.0",
     workspaceRoot: options2.cwd,
     capabilities: {
@@ -476299,7 +476343,7 @@ function getAllReleaseNotes(changelogContent = getStoredChangelogFromMemory()) {
     return [];
   }
 }
-async function checkForReleaseNotes(lastSeenVersion, currentVersion = "1.80.6") {
+async function checkForReleaseNotes(lastSeenVersion, currentVersion = "1.80.8") {
   if (process.env.USER_TYPE === "ant") {
     const changelog = "";
     if (changelog) {
@@ -476326,7 +476370,7 @@ async function checkForReleaseNotes(lastSeenVersion, currentVersion = "1.80.6")
     releaseNotes
   };
 }
-function checkForReleaseNotesSync(lastSeenVersion, currentVersion = "1.80.6") {
+function checkForReleaseNotesSync(lastSeenVersion, currentVersion = "1.80.8") {
   if (process.env.USER_TYPE === "ant") {
     const changelog = "";
     if (changelog) {
@@ -479231,7 +479275,7 @@ function getRecentActivitySync() {
   return cachedActivity;
 }
 function getLogoDisplayData() {
-  const version2 = process.env.DEMO_VERSION ?? "1.80.6";
+  const version2 = process.env.DEMO_VERSION ?? "1.80.8";
   const serverUrl = getDirectConnectServerUrl();
   const displayPath = process.env.DEMO_VERSION ? "/code/ur" : getDisplayPath(getCwd());
   const cwd2 = serverUrl ? `${displayPath} in ${serverUrl.replace(/^https?:\/\//, "")}` : displayPath;
@@ -480099,7 +480143,7 @@ function LogoV2() {
   if ($2[2] === Symbol.for("react.memo_cache_sentinel")) {
     t2 = () => {
       const currentConfig = getGlobalConfig();
-      if (currentConfig.lastReleaseNotesSeen === "1.80.6") {
+      if (currentConfig.lastReleaseNotesSeen === "1.80.8") {
         return;
       }
       saveGlobalConfig(_temp325);
@@ -480784,12 +480828,12 @@ function LogoV2() {
   return t41;
 }
 function _temp325(current) {
-  if (current.lastReleaseNotesSeen === "1.80.6") {
+  if (current.lastReleaseNotesSeen === "1.80.8") {
     return current;
   }
   return {
     ...current,
-    lastReleaseNotesSeen: "1.80.6"
+    lastReleaseNotesSeen: "1.80.8"
   };
 }
 function _temp241(s_0) {
@@ -496881,7 +496925,7 @@ function compileAgenticCiWorkflow(specName = "default", options2 = {}) {
   if (spec.name !== specName) {
     throw new Error("Agentic CI workflow spec name does not match");
   }
-  const packageVersion = options2.packageVersion ?? (typeof MACRO !== "undefined" ? "1.80.6" : "1.80.6");
+  const packageVersion = options2.packageVersion ?? (typeof MACRO !== "undefined" ? "1.80.8" : "1.80.8");
   if (!/^\d+\.\d+\.\d+(?:-[0-9A-Za-z.-]+)?$/.test(packageVersion)) {
     throw new Error("invalid ur-agent package version");
   }
@@ -497874,7 +497918,7 @@ runner; \`ur trigger\` is the inbound parser that decides what to run.
       path: ".github/workflows/ur.yml",
       root: "project",
       content: compileAgenticCiWorkflow("default", {
-        packageVersion: typeof MACRO !== "undefined" ? "1.80.6" : "1.80.6"
+        packageVersion: typeof MACRO !== "undefined" ? "1.80.8" : "1.80.8"
       })
     },
     {
@@ -497937,7 +497981,7 @@ function value(tokens, flag) {
   return index2 >= 0 ? tokens[index2 + 1] : undefined;
 }
 function cliVersion() {
-  return typeof MACRO !== "undefined" ? "1.80.6" : "1.80.6";
+  return typeof MACRO !== "undefined" ? "1.80.8" : "1.80.8";
 }
 function workflowPath(cwd2) {
   return join168(cwd2, ".github", "workflows", "ur-agentic-ci.yml");
@@ -503793,7 +503837,7 @@ function createAcpStdioApp(deps) {
       }
     },
     authMethods: [],
-    agentInfo: { name: "UR-Nexus", version: "1.80.6" }
+    agentInfo: { name: "UR-Nexus", version: "1.80.8" }
   })).onRequest("authenticate", () => ({})).onRequest("session/new", async (context6) => {
     const result = runtime2.newSession(context6.params.cwd, context6.params.mcpServers, context6.params.additionalDirectories);
     await runtime2.announce({
@@ -503890,7 +503934,7 @@ function createAcpStdioAgent(deps) {
               }
             },
             authMethods: [],
-            agentInfo: { name: "UR-Nexus", version: "1.80.6" }
+            agentInfo: { name: "UR-Nexus", version: "1.80.8" }
           });
           return;
         case "authenticate":
@@ -715116,7 +715160,7 @@ async function captureMemoryDiagnostics(trigger2, dumpNumber = 0) {
     smapsRollup,
     platform: process.platform,
     nodeVersion: process.version,
-    ccVersion: "1.80.6"
+    ccVersion: "1.80.8"
   };
 }
 async function performHeapDump(trigger2 = "manual", dumpNumber = 0) {
@@ -715705,7 +715749,7 @@ var init_bridge_kick = __esm(() => {
 var call154 = async () => {
   return {
     type: "text",
-    value: "1.80.6"
+    value: "1.80.8"
   };
 }, version2, version_default;
 var init_version = __esm(() => {
@@ -726948,7 +726992,7 @@ function generateHtmlReport(data, insights) {
 </html>`;
 }
 function buildExportData(data, insights, facets, remoteStats) {
-  const version3 = typeof MACRO !== "undefined" ? "1.80.6" : "unknown";
+  const version3 = typeof MACRO !== "undefined" ? "1.80.8" : "unknown";
   const remote_hosts_collected = remoteStats?.hosts.filter((h2) => h2.sessionCount > 0).map((h2) => h2.name);
   const facets_summary = {
     total: facets.size,
@@ -731261,7 +731305,7 @@ var init_sessionStorage = __esm(() => {
   init_settings2();
   init_slowOperations();
   init_uuid();
-  VERSION7 = typeof MACRO !== "undefined" ? "1.80.6" : "unknown";
+  VERSION7 = typeof MACRO !== "undefined" ? "1.80.8" : "unknown";
   MAX_TOMBSTONE_REWRITE_BYTES = 50 * 1024 * 1024;
   SKIP_FIRST_PROMPT_PATTERN = /^(?:\s*<[a-z][\w-]*[\s>]|\[Request interrupted by user[^\]]*\])/;
   EPHEMERAL_PROGRESS_TYPES = new Set([
@@ -732476,7 +732520,7 @@ var init_filesystem = __esm(() => {
   });
   getBundledSkillsRoot = memoize_default(function getBundledSkillsRoot2() {
     const nonce = randomBytes24(16).toString("hex");
-    return join243(getURTempDir(), "bundled-skills", "1.80.6", nonce);
+    return join243(getURTempDir(), "bundled-skills", "1.80.8", nonce);
   });
   getResolvedWorkingDirPaths = memoize_default(getPathsForPermissionCheck);
 });
@@ -738086,6 +738130,7 @@ function getOllamaToolDisciplineSection() {
     return null;
   const items = [
     `Use the native structured tool-call interface; never substitute prose, fenced code, XML, or printed arguments for a call. Only use a text fallback when the runtime explicitly says native tools are unavailable and supplies the exact fallback format.`,
+    `Call only tools exposed in the active tool list. If a tool result says a tool is unavailable, do not retry it unchanged: use an available alternative or continue with the useful partial result.`,
     `Use ${FILE_WRITE_TOOL_NAME} or ${FILE_EDIT_TOOL_NAME} for file changes. Batch independent calls in one turn (maximum 8); keep read\u2192decide\u2192write and other dependencies sequential.`,
     `Treat each call as pending until its matching result arrives. Observe that result before continuing, and never claim a file change, command, test, or other action succeeded without a successful result.`,
     `Never emit an empty turn: provide a real tool call, useful user-facing text, or both.`
@@ -738095,7 +738140,7 @@ function getOllamaToolDisciplineSection() {
 }
 function getAgentToolSection() {
   const launch = isForkSubagentEnabled() ? `Calling ${AGENT_TOOL_NAME} without a subagent_type creates a fork, which runs in the background and keeps its raw tool output out of your context. **If you ARE the fork**, execute your bounded assignment directly; do not re-delegate.` : `Use ${AGENT_TOOL_NAME} with the specialized agent whose description best matches each bounded assignment.`;
-  return `${launch} For a large request, first create a bounded task list, then delegate at most one ready independent branch per agent. Launch independent read-only research, audits, or exploration together. Shared-checkout writers, unknown scopes, dependencies, and overlapping file targets must run sequentially. Parallel writers require separate worktrees based on the exact clean starting revision; if the current required state is dirty or unsnapshotted, keep those writers serial in the shared checkout. Keep tiny dependent steps with the parent when delegation overhead is larger than the work. The parent owns task status, integration, and final verification: do not duplicate delegated work, do not mark a delegated task complete from a launch acknowledgement, and do not finish until the returned result and acceptance evidence have been checked.`;
+  return `${launch} For a large request, first create a bounded task list, then delegate at most one ready independent branch per agent. Before that list exists, read-only research, audits, and exploration MUST use subagent_type="Explore" (or "Plan" for planning); never label a read-only worker general-purpose. Every custom, general-purpose, write-capable, nested, team, or worktree delegation requires a ready parent task. Launch independent read-only research, audits, or exploration together. Shared-checkout writers, unknown scopes, dependencies, and overlapping file targets must run sequentially. Parallel writers require separate worktrees based on the exact clean starting revision; if the current required state is dirty or unsnapshotted, keep those writers serial in the shared checkout. Keep tiny dependent steps with the parent when delegation overhead is larger than the work. The parent owns task status, integration, and final verification: do not duplicate delegated work, do not mark a delegated task complete from a launch acknowledgement, and do not finish until the returned result and acceptance evidence have been checked.`;
 }
 function getDiscoverSkillsGuidance() {
   if (false) {}
@@ -738865,7 +738910,7 @@ function computeFingerprint(messageText2, version3) {
 }
 function computeFingerprintFromMessages(messages) {
   const firstMessageText = extractFirstMessageText(messages);
-  return computeFingerprint(firstMessageText, "1.80.6");
+  return computeFingerprint(firstMessageText, "1.80.8");
 }
 var FINGERPRINT_SALT = "59cf53e54c78";
 var init_fingerprint = () => {};
@@ -740790,7 +740835,7 @@ async function sideQuery(opts) {
     betas.push(STRUCTURED_OUTPUTS_BETA_HEADER);
   }
   const messageText2 = extractFirstUserMessageText(messages);
-  const fingerprint2 = computeFingerprint(messageText2, "1.80.6");
+  const fingerprint2 = computeFingerprint(messageText2, "1.80.8");
   const attributionHeader = getAttributionHeader(fingerprint2);
   const systemBlocks = [
     attributionHeader ? { type: "text", text: attributionHeader } : null,
@@ -745624,7 +745669,7 @@ function buildSystemInitMessage(inputs) {
     slash_commands: inputs.commands.filter((c4) => c4.userInvocable !== false).map((c4) => c4.name),
     apiKeySource: getURHQApiKeyWithSource().source,
     betas: getSdkBetas(),
-    ur_version: "1.80.6",
+    ur_version: "1.80.8",
     output_style: outputStyle,
     agents: inputs.agents.map((agent2) => agent2.agentType),
     skills: inputs.skills.filter((s) => s.userInvocable !== false).map((skill2) => skill2.name),
@@ -759460,7 +759505,7 @@ var init_useVoiceEnabled = __esm(() => {
 function getSemverPart(version3) {
   return `${import_semver13.major(version3, { loose: true })}.${import_semver13.minor(version3, { loose: true })}.${import_semver13.patch(version3, { loose: true })}`;
 }
-function useUpdateNotification(updatedVersion, initialVersion = "1.80.6") {
+function useUpdateNotification(updatedVersion, initialVersion = "1.80.8") {
   const [lastNotifiedSemver, setLastNotifiedSemver] = import_react225.useState(() => getSemverPart(initialVersion));
   if (!updatedVersion) {
     return null;
@@ -759509,7 +759554,7 @@ function AutoUpdater({
       return;
     }
     if (false) {}
-    const currentVersion = "1.80.6";
+    const currentVersion = "1.80.8";
     const channel = getInitialSettings()?.autoUpdatesChannel ?? "latest";
     let latestVersion = await getLatestVersion(channel);
     const isDisabled = isAutoUpdaterDisabled();
@@ -759738,12 +759783,12 @@ function NativeAutoUpdater({
     logEvent("tengu_native_auto_updater_start", {});
     try {
       const maxVersion = await getMaxVersion();
-      if (maxVersion && gt("1.80.6", maxVersion)) {
+      if (maxVersion && gt("1.80.8", maxVersion)) {
         const msg = await getMaxVersionMessage();
         setMaxVersionIssue(msg ?? "affects your version");
       }
       const result = await installLatest(channel);
-      const currentVersion = "1.80.6";
+      const currentVersion = "1.80.8";
       const latencyMs = Date.now() - startTime;
       if (result.lockFailed) {
         logEvent("tengu_native_auto_updater_lock_contention", {
@@ -759880,17 +759925,17 @@ function PackageManagerAutoUpdater(t0) {
       const maxVersion = await getMaxVersion();
       if (maxVersion && latest && gt(latest, maxVersion)) {
         logForDebugging(`PackageManagerAutoUpdater: maxVersion ${maxVersion} is set, capping update from ${latest} to ${maxVersion}`);
-        if (gte("1.80.6", maxVersion)) {
-          logForDebugging(`PackageManagerAutoUpdater: current version ${"1.80.6"} is already at or above maxVersion ${maxVersion}, skipping update`);
+        if (gte("1.80.8", maxVersion)) {
+          logForDebugging(`PackageManagerAutoUpdater: current version ${"1.80.8"} is already at or above maxVersion ${maxVersion}, skipping update`);
           setUpdateAvailable(false);
           return;
         }
         latest = maxVersion;
       }
-      const hasUpdate = latest && !gte("1.80.6", latest) && !shouldSkipVersion(latest);
+      const hasUpdate = latest && !gte("1.80.8", latest) && !shouldSkipVersion(latest);
       setUpdateAvailable(!!hasUpdate);
       if (hasUpdate) {
-        logForDebugging(`PackageManagerAutoUpdater: Update available ${"1.80.6"} -> ${latest}`);
+        logForDebugging(`PackageManagerAutoUpdater: Update available ${"1.80.8"} -> ${latest}`);
       }
     };
     $2[0] = t1;
@@ -759924,7 +759969,7 @@ function PackageManagerAutoUpdater(t0) {
       wrap: "truncate",
       children: [
         "currentVersion: ",
-        "1.80.6"
+        "1.80.8"
       ]
     }, undefined, true, undefined, this);
     $2[3] = verbose;
@@ -770777,7 +770822,7 @@ function buildStatusLineCommandInput(permissionMode, exceeds200kTokens, settings
       project_dir: getOriginalCwd(),
       added_dirs: addedDirs
     },
-    version: "1.80.6",
+    version: "1.80.8",
     output_style: {
       name: outputStyleName
     },
@@ -770912,7 +770957,7 @@ function StatusLineInner({
   const attention = customStatusError ?? taskAttention;
   const terminalSize = React133.useContext(TerminalSizeContext);
   const defaultStatusLineText = buildDefaultStatusBar({
-    version: "1.80.6",
+    version: "1.80.8",
     providerLabel: providerRuntime.providerLabel,
     authMode: providerRuntime.authLabel,
     model: renderModelName(mainLoopModel) || providerRuntime.model || "",
@@ -783167,7 +783212,7 @@ async function submitTranscriptShare(messages, trigger2, appearanceId) {
     } catch {}
     const data = {
       trigger: trigger2,
-      version: "1.80.6",
+      version: "1.80.8",
       platform: process.platform,
       transcript,
       subagentTranscripts: Object.keys(subagentTranscripts).length > 0 ? subagentTranscripts : undefined,
@@ -795536,7 +795581,7 @@ function WelcomeV2() {
             dimColor: true,
             children: [
               "v",
-              "1.80.6"
+              "1.80.8"
             ]
           }, undefined, true, undefined, this)
         ]
@@ -796796,7 +796841,7 @@ function completeOnboarding() {
   saveGlobalConfig((current) => ({
     ...current,
     hasCompletedOnboarding: true,
-    lastOnboardingVersion: "1.80.6"
+    lastOnboardingVersion: "1.80.8"
   }));
 }
 function showDialog(root2, renderer) {
@@ -801942,7 +801987,7 @@ function appendToLog(path28, message) {
     cwd: getFsImplementation().cwd(),
     userType: process.env.USER_TYPE,
     sessionId: getSessionId(),
-    version: "1.80.6"
+    version: "1.80.8"
   };
   getLogWriter(path28).write(messageWithTimestamp);
 }
@@ -806106,8 +806151,8 @@ async function getEnvLessBridgeConfig() {
 }
 async function checkEnvLessBridgeMinVersion() {
   const cfg = await getEnvLessBridgeConfig();
-  if (cfg.min_version && lt("1.80.6", cfg.min_version)) {
-    return `Your version of UR (${"1.80.6"}) is too old for Remote Control.
+  if (cfg.min_version && lt("1.80.8", cfg.min_version)) {
+    return `Your version of UR (${"1.80.8"}) is too old for Remote Control.
 Version ${cfg.min_version} or higher is required. Run \`ur update\` to update.`;
   }
   return null;
@@ -806581,7 +806626,7 @@ async function initBridgeCore(params) {
   const rawApi = createBridgeApiClient({
     baseUrl,
     getAccessToken,
-    runnerVersion: "1.80.6",
+    runnerVersion: "1.80.8",
     onDebug: logForDebugging,
     onAuth401,
     getTrustedDeviceToken
@@ -816054,7 +816099,7 @@ function getAgUiCapabilities() {
       name: "UR-Nexus",
       type: "ur-nexus",
       description: "Provider-flexible, local-first autonomous engineering workflow agent.",
-      version: "1.80.6",
+      version: "1.80.8",
       provider: "UR",
       documentationUrl: "https://github.com/Maitham16/UR/blob/master/docs/AG_UI.md"
     },
@@ -817281,7 +817326,7 @@ function createMCPServer(cwd4, debug2, verbose) {
   };
   const server2 = new Server({
     name: "ur-nexus",
-    version: "1.80.6"
+    version: "1.80.8"
   }, {
     capabilities: {
       tools: {}
@@ -818485,7 +818530,7 @@ function thrownResponse(error40) {
 }
 async function createUrMcp2026Runtime(options5) {
   const server2 = createMCPServer(options5.cwd, options5.debug === true, options5.verbose === true);
-  const client2 = new Client({ name: "ur-mcp-2026-adapter", version: "1.80.6" }, { capabilities: {} });
+  const client2 = new Client({ name: "ur-mcp-2026-adapter", version: "1.80.8" }, { capabilities: {} });
   const [clientTransport, serverTransport] = createLinkedTransportPair();
   try {
     await server2.connect(serverTransport);
@@ -818496,7 +818541,7 @@ async function createUrMcp2026Runtime(options5) {
   }
   const runtime2 = new Mcp2026Runtime({
     cwd: options5.cwd,
-    version: "1.80.6",
+    version: "1.80.8",
     backend: {
       listTools: async () => {
         const listed = await client2.listTools();
@@ -821231,7 +821276,7 @@ async function update() {
   logEvent("tengu_update_check", {});
   const diagnostic2 = await getDoctorDiagnostic();
   const result = await checkUpgradeStatus({
-    currentVersion: "1.80.6",
+    currentVersion: "1.80.8",
     packageName: UR_AGENT_PACKAGE_NAME,
     installationType: diagnostic2.installationType,
     latestVersion: () => getLatestNpmPackageVersion(UR_AGENT_PACKAGE_NAME)
@@ -822559,7 +822604,7 @@ ${customInstructions}` : customInstructions;
       }
     }
     logForDiagnosticsNoPII("info", "started", {
-      version: "1.80.6",
+      version: "1.80.8",
       is_native_binary: isInBundledMode()
     });
     registerCleanup(async () => {
@@ -823346,7 +823391,7 @@ Usage: ur --remote "your task description"`, () => gracefulShutdown(1));
         pendingHookMessages
       }, renderAndRun);
     }
-  }).version("1.80.6 (UR-Nexus)", "-v, --version", "Output the version number");
+  }).version("1.80.8 (UR-Nexus)", "-v, --version", "Output the version number");
   program2.option("-w, --worktree [name]", "Create a new git worktree for this session (optionally specify a name)");
   program2.option("--tmux", "Create a tmux session for the worktree (requires --worktree). Uses iTerm2 native panes when available; use --tmux=classic for traditional tmux.");
   if (canUserConfigureAdvisor()) {
@@ -824473,7 +824518,7 @@ if (false) {}
 async function main2() {
   const args = process.argv.slice(2);
   if (args.length === 1 && (args[0] === "--version" || args[0] === "-v" || args[0] === "-V")) {
-    console.log(`${"1.80.6"} (UR-Nexus)`);
+    console.log(`${"1.80.8"} (UR-Nexus)`);
     return;
   }
   if (args[0] === "a2a" && args[1] === "serve" && !args.includes("--help") && !args.includes("-h")) {

package/docs/AGENT_FEATURES.md

@@ -9,6 +9,19 @@ reproducible autonomous software engineering agent: every substantial task can
 be driven as `spec -> plan -> patch -> test -> report -> rollback`, with the
 spec as the durable source of truth and command evidence as the success gate.
 
+## v1.80.8 Addition
+
+| Addition | Surface | What it adds |
+| --- | --- | --- |
+| Provider-independent research routing | Shared Agent prompt and execution gate | Directs every model family to use the shipped `Explore` worker for task-free read-only research. If a model still labels an explicitly read-only research brief `general-purpose`, UR reduces it to protected Explore capabilities before task gating; write-capable and custom delegation remains gated. |
+
+## v1.80.7 Additions
+
+| Addition | Surface | What it adds |
+| --- | --- | --- |
+| Recoverable local-model tool mismatch | Ollama native/text calls, streaming/non-streaming execution | Converts a valid call to a tool absent from the active profile into a safe `UnavailableTool` result instead of aborting the provider turn. Identical retries are bounded and omitted tools cannot be revived through legacy aliases. |
+| Task-free read-only research | `Agent`, strict-hybrid task gate | Lets the main session launch the exact shipped `Explore` and `Plan` definitions before tasks exist in every parent permission mode, while forcing those children into plan permissions and keeping all custom or write-capable delegation gated. |
+
 ## v1.80.6 Addition
 
 | Addition | Surface | What it adds |

package/docs/CONFIGURATION.md

@@ -56,7 +56,14 @@ never blocked.
 do not carry a classified user turn; it is not a limit on investigation. Set
 it to `0` to require a task before every mutation. Set `enabled` to `false` to
 return to advisory task tracking. Profiles that omit `TaskCreate` are not
-gated, because they could not satisfy the requirement.
+gated, because they could not satisfy the requirement. The main session may
+launch UR's shipped `Explore` and `Plan` agents before a task exists in any
+permission mode. These exact built-in definitions are forced read-only; custom,
+write-capable, nested, named/team, and worktree delegation still requires an
+actionable parent task. As provider-independent compatibility, an unnamed
+main-session `general-purpose` call whose own prompt explicitly declares both
+research intent and no file writes is reduced to the shipped Explore definition
+before the gate. It never receives general-purpose tools through this path.
 
 ## Model Providers
 

package/docs/TROUBLESHOOTING.md

@@ -83,14 +83,17 @@ ur provider status
   different source. Timeouts, `408`, `409`, `425`, `429`, and `5xx` responses
   remain retryable because they may be transient.
 
-### Plan mode says `TaskListRequired`
+### A built-in research agent says `TaskListRequired`
 
 - Likely cause on UR 1.80.3–1.80.5: the strict task gate treated either the
   active plan file or early read-only research delegation as an untracked
   project mutation, creating a circular requirement before planning finished.
-- Fix: upgrade to UR 1.80.6 or newer. The exact active plan file and main-thread
-  delegation to UR's shipped read-only `Explore` and `Plan` agents are allowed
-  before tasks exist. Approved plans are synchronized into visible
+- UR 1.80.6 fixed this inside Plan Mode. UR 1.80.7 extends the same safe rule to
+  ordinary main-session research: upgrade to 1.80.7 or newer. The exact active
+  plan file and main-thread delegation to UR's shipped read-only `Explore` and
+  `Plan` agents are allowed before tasks exist. The child is forced into plan
+  permissions even if the parent uses Accept Edits or Approve All. Approved
+  plans are synchronized into visible
   implementation and verification tasks before the first project mutation.
   Other files and write-capable delegation remain protected. Existing
   actionable tasks are preserved.
@@ -99,6 +102,26 @@ ur provider status
   agents without a parent task: create the requested tasks or finish and
   approve the plan first. Disabling `tasks.requireBeforeChanges` is no longer
   needed for normal plan mode.
+- If an API, local, or subscription-CLI model still returns this error for an
+  explicitly read-only research brief on 1.80.7, upgrade to 1.80.8. Some models
+  emitted `subagent_type: "general-purpose"` despite the read-only instruction.
+  UR 1.80.8 routes that narrow contract to the protected Explore worker before
+  the gate. A real general-purpose or implementation worker is still expected
+  to require a task.
+
+### Ollama stops with `unavailable tool "WebSearch"`
+
+- Cause on UR 1.80.6 and older: a local model requested a provider-hosted tool
+  that was not present in its active tool profile. The Ollama adapter treated
+  the valid but unavailable tool name as a fatal provider-response error, so
+  the parent could not receive the research agents' remaining useful results.
+- Fix: upgrade to UR 1.80.7 or newer. UR now returns a recoverable
+  `UnavailableTool` result without executing the call. The agent is instructed
+  to use an available alternative or return its partial result, and repeated
+  identical unavailable calls are bounded. This applies to native and
+  text-form calls in streaming and non-streaming Ollama responses.
+- `WebSearch` is still not fabricated for a model or profile that does not have
+  it. Malformed tool names and malformed arguments still fail closed.
 
 ## Providers and models
 

package/docs/USAGE.md

@@ -470,11 +470,15 @@ older builds are hidden and removed at the next prompt boundary.
 
 Plan mode has one narrow exception to the mutation gate: UR may write or edit
 the exact plan file for the active session while the rest of the workspace
-remains read-only. The main session may also delegate early research to UR's
-shipped `Explore` and `Plan` agents before tasks exist. Those two definitions
-are mechanically held in plan permission mode; custom overrides,
-general-purpose agents, nested agents, team workers, and worktree agents still
-require an actionable parent task. When the user approves the plan,
+remains read-only. In any permission mode, the main session may delegate early
+research to UR's shipped `Explore` and `Plan` agents before tasks exist. Those
+two definitions are mechanically forced into plan permission mode even when
+the parent is in Accept Edits or Approve All. Models are instructed to select
+`subagent_type="Explore"` for this work. If a provider still labels an explicit
+read-only research brief `general-purpose`, UR reduces that main-session call
+to the shipped Explore definition before gating. Ordinary general-purpose
+work, custom overrides, nested agents, named/team workers, worktrees, and cwd
+overrides still require an actionable parent task. When the user approves the plan,
 `ExitPlanMode` preserves any existing actionable board or creates a bounded set
 of professional, deduplicated implementation tasks plus a verification task
 that depends on them. Implementation therefore starts with visible tracking

package/docs/VALIDATION.md

@@ -19,22 +19,43 @@ You need:
 
 ```sh
 ur --version
-# expected for this release: "1.80.6 (UR-Nexus)"
+# expected for this release: "1.80.8 (UR-Nexus)"
 ```
 
-### 0.0 Plan research delegation starts cleanly (1.80.6)
+### 0.0 Read-only research delegation starts cleanly (1.80.8)
 
-Start an interactive session with task enforcement enabled and ask UR to plan a
-change that requires codebase research. Before any task exists, built-in
-`Explore` and `Plan` agent calls should initialize without `TaskListRequired`.
-Custom or general-purpose agents should remain blocked until they have an
-actionable parent task. The deterministic regressions are:
+Start an interactive session with task enforcement enabled and ask UR to
+research a change, both normally and from Plan Mode. Before any task exists,
+built-in `Explore` and `Plan` agent calls should initialize without
+`TaskListRequired`. Their workers remain read-only even when the parent uses
+Accept Edits or Approve All. Custom agents and ordinary general-purpose agents
+should remain blocked until they have an actionable parent task.
+
+Repeat with multiple available model families. A model may correctly emit
+`subagent_type: "Explore"`; if it instead emits `general-purpose` with an
+explicit read-only research/no-file-write brief, UR must safely reduce it to
+Explore and start without a failed `TaskListRequired` attempt. A
+general-purpose implementation brief must remain blocked. The deterministic
+regressions are:
 
 ```sh
 bun test test/taskListGate.test.ts test/toolExecutionFinalInput.test.ts
 ```
 
-### 0.0.1 Plan approval creates visible tasks (1.80.5)
+### 0.0.1 Unavailable Ollama tools recover (1.80.7)
+
+With an Ollama model, ask for research that mentions WebSearch. If WebSearch is
+not in the active profile, UR should reject that call safely and the agent
+should continue with available tools or its useful partial result. It must not
+end the parent turn with `Ollama response returned unavailable tool`. Native
+and text-form, streaming and non-streaming regressions are covered by:
+
+```sh
+bun test test/ollamaToolCalls.test.ts test/kimiToolCalls.test.ts \
+  test/repeatedFailureGuard.test.ts test/streamingToolExecutor.test.ts
+```
+
+### 0.0.2 Plan approval creates visible tasks (1.80.5)
 
 Start an interactive session with task enforcement enabled, ask for a
 multi-file change, and let the agent enter plan mode. Expected lifecycle:

package/documentation/index.html

@@ -45,7 +45,7 @@
     <main id="content" class="content">
       <header class="topbar">
         <div>
-          <p class="eyebrow">Version 1.80.6</p>
+          <p class="eyebrow">Version 1.80.8</p>
           <h1>UR-Nexus Documentation</h1>
           <p class="lead">A practical, tutorial-style reference for installing, configuring, automating, extending, and operating UR-Nexus.</p>
         </div>

package/extensions/jetbrains-ur/build.gradle.kts

@@ -7,7 +7,7 @@ plugins {
 }
 
 group = "dev.urnexus"
-version = "1.80.6"
+version = "1.80.8"
 
 repositories {
     mavenCentral()

package/extensions/vscode-ur-inline-diffs/package.json

@@ -2,7 +2,7 @@
   "name": "ur-inline-diffs",
   "displayName": "UR Inline Diffs",
   "description": "Review, apply, and reject UR inline diff bundles from .ur/ide/diffs inside VS Code.",
-  "version": "1.80.6",
+  "version": "1.80.8",
   "publisher": "ur-nexus",
   "engines": {
     "vscode": "^1.92.0"

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ur-agent",
-  "version": "1.80.6",
+  "version": "1.80.8",
   "description": "UR-Nexus — autonomous engineering workflow engine (plan, execute, test, verify, document, benchmark, reproduce)",
   "type": "module",
   "packageManager": "bun@1.3.14",