explorbot 0.1.25 → 0.1.27

package/dist/package.json

@@ -1,6 +1,6 @@
 {
   "name": "explorbot",
-  "version": "0.1.25",
+  "version": "0.1.27",
   "description": "CLI app built with React Ink, CodeceptJS, and Playwright",
   "license": "Elastic-2.0",
   "type": "module",

package/dist/src/ai/historian/codeceptjs.js

@@ -82,7 +82,7 @@ export function WithCodeceptJS(Base) {
             const filePath = join(testsDir, safeFilename(plan.title, '.js'));
             writeFileSync(filePath, lines.join('\n'));
             this.savedFiles.add(filePath);
-            tag('substep').log(`Saved plan tests to: ${relativeToCwd(filePath)}`);
+            tag('operation').log(`Saved plan tests to: ${relativeToCwd(filePath)}`);
             return filePath;
         }
         getKnowledgeLines(url, indent = '  ') {

package/dist/src/ai/historian/experience.js

@@ -30,7 +30,7 @@ export function WithExperience(Base) {
                 await this.reportSession(task, steps);
             }
             await this.stopScreencast();
-            tag('substep').log(`Historian saved session for: ${task.description}`);
+            tag('operation').log(`Historian saved session for: ${task.description}`);
         }
         async reportSession(test, steps) {
             if (!this.reporter)

package/dist/src/ai/historian/playwright.js

@@ -119,7 +119,7 @@ export function WithPlaywright(Base) {
             const filePath = join(testsDir, safeFilename(plan.title, '.spec.ts'));
             writeFileSync(filePath, lines.join('\n'));
             this.savedFiles.add(filePath);
-            tag('substep').log(`Saved plan tests to: ${relativeToCwd(filePath)}`);
+            tag('operation').log(`Saved plan tests to: ${relativeToCwd(filePath)}`);
             return filePath;
         }
         getPlaywrightKnowledgeLines(url, indent = '    ') {

package/dist/src/ai/historian/screencast.js

@@ -76,7 +76,7 @@ export function WithScreencast(Base) {
                 this.screencastLastChapter = null;
             }
             catch (err) {
-                tag('substep').log(`Screencast start failed: ${err.message}`);
+                tag('operation').log(`Screencast start failed: ${err.message}`);
             }
         }
         async emitChapter(_step) {
@@ -104,7 +104,7 @@ export function WithScreencast(Base) {
                 await this.screencastPage.screencast.stop();
             }
             catch (err) {
-                tag('substep').log(`Screencast stop failed: ${err.message}`);
+                tag('operation').log(`Screencast stop failed: ${err.message}`);
             }
             this.screencastActive = false;
             this.screencastPage = null;
@@ -114,7 +114,7 @@ export function WithScreencast(Base) {
             if (path) {
                 this.savedFiles.add(path);
                 task?.addArtifact?.(path);
-                tag('substep').log(`Saved screencast: ${relativeToCwd(path)}`);
+                tag('operation').log(`Saved screencast: ${relativeToCwd(path)}`);
             }
         }
     };

package/dist/src/ai/historian.js

@@ -38,6 +38,6 @@ export class Historian extends HistorianBase {
         }
         writeFileSync(filePath, content);
         this.savedFiles.add(filePath);
-        tag('substep').log(`Updated test file with healed steps: ${relativeToCwd(filePath)}`);
+        tag('operation').log(`Updated test file with healed steps: ${relativeToCwd(filePath)}`);
     }
 }

package/dist/src/ai/navigator.js

@@ -182,7 +182,7 @@ class Navigator {
         if (!actionResult.isInsideIframe) {
             const successful = this.experienceTracker.getSuccessfulExperience(actionResult);
             if (successful.length > 0) {
-                tag('substep').log(`Found ${successful.length} experience ${pluralize(successful.length, 'file')} for: ${actionResult.url}`);
+                tag('operation').log(`Found ${successful.length} experience ${pluralize(successful.length, 'file')} for: ${actionResult.url}`);
                 experience = `<experience>\nPast successful recipes recorded from prior runs for this page. Prefer these solutions first if they match the goal.\n\n${successful.join('\n\n')}\n</experience>`;
             }
         }
@@ -276,7 +276,7 @@ class Navigator {
                     stop();
                     return;
                 }
-                tag('substep').log('Feeding failures back to AI for a new batch...');
+                tag('operation').log('Feeding failures back to AI for a new batch...');
                 let contextMsg = 'Previous solutions did not work. Analyze the failures and try DIFFERENT strategies (not syntactic variants of the same locator).\n\n';
                 if (batchFailures.length > 0) {
                     const lines = batchFailures
@@ -573,7 +573,7 @@ class Navigator {
         debugLog('Verification message:', message);
         const cachedVerification = actionResult.getVerification(message);
         if (cachedVerification !== null) {
-            tag('substep').log(`Reusing cached verification: ${cachedVerification ? 'PASS' : 'FAIL'}`);
+            tag('operation').log(`Reusing cached verification: ${cachedVerification ? 'PASS' : 'FAIL'}`);
             return { verified: cachedVerification, successfulCodes: [], assertionSteps: [], totalAttempted: 0 };
         }
         let knowledge = '';
@@ -591,7 +591,7 @@ class Navigator {
             const toc = this.experienceTracker.getExperienceTableOfContents(actionResult);
             if (toc.length > 0) {
                 const totalSections = toc.reduce((sum, entry) => sum + entry.sections.length, 0);
-                tag('substep').log(`Found ${toc.length} experience ${pluralize(toc.length, 'file')} (${totalSections} sections) for: ${actionResult.url}`);
+                tag('operation').log(`Found ${toc.length} experience ${pluralize(toc.length, 'file')} (${totalSections} sections) for: ${actionResult.url}`);
                 experience = renderExperienceToc(toc);
             }
         }

package/dist/src/ai/pilot.js

@@ -84,7 +84,7 @@ export class Pilot {
         }
         const schema = z.object({
             decision: z.enum(['pass', 'fail', 'continue', 'skipped']).describe('pass = test succeeded, fail = test failed, continue = tester should keep going, skipped = scenario is irrelevant OR systematic execution failures prevented testing'),
-            reason: z.string().describe('What happened and why (1-2 sentences). Do NOT repeat the decision status (e.g. "scenario goal achieved/not achieved") — just explain the evidence. For continue: explain why rejected and suggest alternatives.'),
+            reason: z.string().describe('Concise user-facing reason, maximum 1 short sentence and 120 characters. Do NOT repeat the decision status; explain only the evidence. For continue: explain why rejected and suggest alternatives.'),
             guidance: z.string().nullable().describe('Required for "continue": specific actionable instruction for the tester — what exactly to verify, retry differently, or complete next. Be concrete.'),
             requestVerification: z
                 .string()
@@ -149,7 +149,7 @@ export class Pilot {
                     this.explorer.getPlaywrightRecorder().recordVerification(verifyResult.assertionSteps);
                 }
             }
-            tag('info').log(`Pilot: ${result.decision} — ${result.reason}`);
+            tag('info').log(`Pilot: ${result.decision} - ${result.reason}`);
             task.summary = result.reason;
             const verdictState = screenshotState || currentState;
             if (result.decision === 'pass') {
@@ -187,7 +187,7 @@ export class Pilot {
         const notes = task.notesToString() || 'No notes recorded.';
         const schema = z.object({
             decision: z.enum(['allow', 'fail', 'continue', 'skipped']).describe('allow = reset proceeds, fail = test failed (stop looping), continue = veto reset, tester should act on current page instead, skipped = scenario is irrelevant or cannot be executed'),
-            reason: z.string().describe('What evidence justifies this decision (1-2 sentences). Do not restate the decision.'),
+            reason: z.string().describe('Concise evidence-only reason, maximum 1 short sentence and 120 characters. Do not restate the decision.'),
             guidance: z.string().nullable().describe('Required for "continue": concrete instruction for what the tester should do instead of resetting (e.g. which tool to call, what to verify).'),
         });
         const userContent = dedent `
@@ -343,8 +343,9 @@ export class Pilot {
       - "continue": tester hasn't completed the goal; provide concrete guidance (which tool, what to check).
         If a verify() asserted a state that was ALREADY TRUE before the test, it proves nothing — reject.
 
-      reason field: do NOT restate the decision ("scenario goal achieved/not achieved"). State what happened —
-      what was verified, what failed, what evidence was found.
+      reason field: one short sentence, maximum 120 characters. Do NOT restate the decision
+      ("scenario goal achieved/not achieved"). State what happened: what was verified, what failed,
+      or what evidence was found.
     `;
     }
     async planTest(task, currentState) {
@@ -919,6 +920,8 @@ export class Pilot {
       Response format:
       PROGRESS: <1 sentence assessment>
       NEXT: <specific actionable instruction for Tester>
+
+      Keep user-facing reasons concise: one short sentence, maximum 120 characters, evidence only, no repeated verdict wording.
     `;
     }
 }

package/dist/src/ai/planner.js

@@ -14,7 +14,7 @@ import { mdq } from '../utils/markdown-query.js';
 import { planToCompactAiContext } from "../utils/test-plan-markdown.js";
 import { Conversation } from "./conversation.js";
 import { WithSessionDedup } from "./planner/session-dedup.js";
-import { getActiveStyle, getStyles } from "./planner/styles.js";
+import { getActiveStyle } from "./planner/styles.js";
 import { WithSubPages, getPlannedByStateHash, getRegisteredPlan, registerPlan } from "./planner/subpages.js";
 import { POSSIBLE_SECTIONS, Researcher } from "./researcher.js";
 import { findSimilarStateHash } from "./researcher/cache.js";
@@ -28,7 +28,7 @@ const TasksSchema = z.object({
         scenario: z.string().describe('A single sentence describing what to test'),
         priority: z.enum(['critical', 'important', 'high', 'normal', 'low']).describe('Priority of the task based on business importance'),
         startUrl: z.string().nullable().describe('Start URL for the test if different from plan URL (only for tests on visited subpages)'),
-        steps: z.array(z.string()).describe('List of steps to perform for this scenario. Each step should be a specific action (e.g., "Click on Login button", "Enter username in email field", "Submit the form"). Keep steps atomic and actionable.'),
+        steps: z.array(z.string()).describe('List of steps to perform for this scenario. Each step should be a specific action (e.g., "Open the form", "Enter required data", "Submit the form"). Keep steps atomic and actionable.'),
         expectedOutcomes: z
             .array(z.string())
             .describe('List of expected outcomes that can be verified. Each outcome should be simple, specific, and easy to check (e.g., "Success message appears", "URL changes to /dashboard", "Form field shows error"). Keep outcomes atomic - do not combine multiple checks into one.'),
@@ -199,9 +199,7 @@ export class Planner extends PlannerBase {
                 tag('multiline').log(summary);
             }
         }
-        const availableStyles = Object.keys(getStyles()).join(', ');
         tag('success').log(`Planning complete! ${this.currentPlan.tests.length} tests in plan: ${this.currentPlan.title}`);
-        tag('info').log(`Planning style: ${this.lastStyleName} (available: ${availableStyles})`);
         if (state.url)
             registerPlan(state.url, this.currentPlan, feature, state.hash);
         this.registerPlanInSession(this.currentPlan);

package/dist/src/ai/provider.js

@@ -259,8 +259,11 @@ export class Provider {
                 }
                 throw new ContextLengthError(error.message || error.toString());
             }
-            tag('error').log(error.message || error.toString());
-            throw new AiError(error.message || error.toString());
+            const message = error.message || error.toString();
+            if (message !== 'No response text from AI') {
+                tag('error').log(message);
+            }
+            throw new AiError(message);
         }
     }
     async generateWithTools(messages, model, tools, options = {}) {
@@ -333,7 +336,6 @@ export class Provider {
         catch (error) {
             clearActivity();
             if (error?.message?.includes('Tool choice is required')) {
-                tag('warning').log('Model completed without calling a tool, returning empty result');
                 return { text: '', toolCalls: [], toolResults: [], response: { messages: [] }, usage: null };
             }
             if (error?.name === 'AbortError')

package/dist/src/ai/researcher/cache.js

@@ -46,6 +46,14 @@ export function getCachedResearch(hash) {
     memoryCacheTimestamps[hash] = now;
     return cached;
 }
+export function getPreviousResearch(hash) {
+    if (!hash)
+        return '';
+    const researchFile = outputPath('research', `${hash}.md`);
+    if (!existsSync(researchFile))
+        return '';
+    return readFileSync(researchFile, 'utf8');
+}
 export function saveResearch(hash, text, combinedHtml) {
     const researchDir = outputPath('research');
     const researchFile = join(researchDir, `${hash}.md`);

package/dist/src/ai/researcher/deep-analysis.js

@@ -2,9 +2,10 @@ import dedent from 'dedent';
 import { ActionResult } from '../../action-result.js';
 import { executionController } from "../../execution-controller.js";
 import { detectFocusArea, diffAriaSnapshots } from "../../utils/aria.js";
+import { extractCodeBlocks } from "../../utils/code-extractor.js";
 import { tag } from '../../utils/logger.js';
 import { mdq } from "../../utils/markdown-query.js";
-import { getCachedResearch, saveResearch } from "./cache.js";
+import { getCachedResearch, getPreviousResearch, saveResearch } from "./cache.js";
 import { debugLog } from "./mixin.js";
 import { parseResearchSections } from "./parser.js";
 const DEFAULT_MAX_EXPANDABLE_CLICKS = 10;
@@ -13,13 +14,26 @@ export function WithDeepAnalysis(Base) {
         async performDeepAnalysis(state, result) {
             tag('info').log('Starting deep analysis of expandable elements');
             await this.navigateTo(state.fullUrl || state.url);
-            let expandables = await this._discoverExpandables(result.text);
-            if (expandables.length === 0) {
-                tag('info').log('No expandable elements identified by AI');
-                return;
-            }
-            tag('substep').log(`Identified ${expandables.length} expandable elements`);
             const maxClicks = this.explorer.getConfig().ai?.agents?.researcher?.maxExpandableClicks ?? DEFAULT_MAX_EXPANDABLE_CLICKS;
+            const expandedSections = [];
+            const navigationLinks = [];
+            let verifiedCodes = [];
+            let missing = [];
+            const previousSections = this._loadPreviousExtendedSections(state.hash || '');
+            if (previousSections.length > 0) {
+                tag('substep').log(`Replaying ${previousSections.length} previously discovered sections`);
+                const replay = await this._replayPreviousSections(state, previousSections, maxClicks);
+                expandedSections.push(...replay.verified);
+                verifiedCodes = replay.verifiedCodes;
+                missing = replay.missing;
+                tag('info').log(`Reused ${replay.verified.length}/${previousSections.length} previous sections, ${missing.length} to re-discover`);
+                if (missing.length === 0 && replay.verified.length >= maxClicks) {
+                    tag('info').log('Page appears unchanged, reusing previous sections and skipping discovery');
+                    this._appendExtendedResearch(result, expandedSections, navigationLinks);
+                    return;
+                }
+            }
+            let expandables = await this._discoverExpandables(result.text, missing, verifiedCodes);
             if (expandables.length > maxClicks) {
                 expandables = await this._selectExpandables(expandables, state.fullUrl || state.url, maxClicks);
                 tag('substep').log(`Selected ${expandables.length} expandables to click (max: ${maxClicks})`);
@@ -29,9 +43,11 @@ export function WithDeepAnalysis(Base) {
                 commands: this._buildClickCommands(el),
                 description: el.name,
             }))
-                .filter((el) => el.commands.length > 0);
+                .filter((el) => el.commands.length > 0)
+                .filter((el) => !el.commands.some((cmd) => verifiedCodes.includes(cmd)));
             if (elements.length === 0) {
-                tag('info').log('No expandables with valid locators');
+                tag('info').log('No new expandable elements to click');
+                this._appendExtendedResearch(result, expandedSections, navigationLinks);
                 return;
             }
             const expandableRows = elements.map((el) => `| ${el.description} | \`${el.commands[0]}\` |`).join('\n');
@@ -39,21 +55,9 @@ export function WithDeepAnalysis(Base) {
             for (const el of elements)
                 debugLog(`Expandable: ${el.description} → ${el.commands[0]}`);
             tag('substep').log(`Clicking ${elements.length} expandable elements`);
-            const expandedSections = [];
-            const navigationLinks = [];
             await this._clickExpandableElements(elements, state, expandedSections, navigationLinks);
             tag('info').log(`Deep analysis complete. Sections: ${expandedSections.length}, navigation links: ${navigationLinks.length}`);
-            const dedupedSections = this._deduplicateExpandedSections(expandedSections);
-            if (dedupedSections.length !== expandedSections.length) {
-                tag('substep').log(`Deduplicated ${expandedSections.length} → ${dedupedSections.length} extended sections`);
-            }
-            if (dedupedSections.length > 0) {
-                result.text += `\n\n# Extended Research\n\n${dedupedSections.join('\n\n---\n\n')}`;
-            }
-            if (navigationLinks.length > 0) {
-                const links = navigationLinks.map((l) => `- \`${l.code}\` opens ${l.url}`).join('\n');
-                result.text += `\n\n## Navigation Links\n\n${links}`;
-            }
+            this._appendExtendedResearch(result, expandedSections, navigationLinks);
         }
         async researchOverlay(current, previous, pageStateHash) {
             const focusArea = detectFocusArea(current.ariaSnapshot);
@@ -97,7 +101,66 @@ export function WithDeepAnalysis(Base) {
             tag('substep').log(`Overlay research appended: ${focusArea.name}`);
             return sectionMarkdown;
         }
-        async _discoverExpandables(researchText) {
+        _loadPreviousExtendedSections(hash) {
+            if (!hash)
+                return [];
+            const previous = getPreviousResearch(hash);
+            if (!previous)
+                return [];
+            const sections = [];
+            for (const section of parseResearchSections(previous)) {
+                if (!section.isExtended)
+                    continue;
+                const code = extractCodeBlocks(section.rawMarkdown)[0];
+                if (!code)
+                    continue;
+                sections.push({ name: section.name, code });
+            }
+            return sections;
+        }
+        async _replayPreviousSections(state, prevSections, maxClicks) {
+            const originalAria = state.ariaSnapshot || '';
+            const verified = [];
+            const verifiedCodes = [];
+            const missing = [];
+            for (const section of prevSections.slice(0, maxClicks)) {
+                if (executionController.isInterrupted())
+                    break;
+                let outcome;
+                try {
+                    outcome = await this._executeAndAnalyze([section.code], section.name, state, originalAria, this._summarizeExpanded(verified));
+                }
+                catch (err) {
+                    tag('warning').log(`Replay failed for "${section.name}": ${err instanceof Error ? err.message : err}`);
+                    await this._restorePageState(state.url, originalAria).catch(() => { });
+                    missing.push(section);
+                    continue;
+                }
+                if (outcome.status === 'revealed') {
+                    verified.push(outcome.sectionMarkdown);
+                    verifiedCodes.push(section.code);
+                    debugLog(`Replayed and verified section: ${section.name}`);
+                    continue;
+                }
+                debugLog(`Could not replay previous section: ${section.name}`);
+                missing.push(section);
+            }
+            return { verified, verifiedCodes, missing };
+        }
+        _appendExtendedResearch(result, expandedSections, navigationLinks) {
+            const dedupedSections = this._deduplicateExpandedSections(expandedSections);
+            if (dedupedSections.length !== expandedSections.length) {
+                tag('substep').log(`Deduplicated ${expandedSections.length} → ${dedupedSections.length} extended sections`);
+            }
+            if (dedupedSections.length > 0) {
+                result.text += `\n\n# Extended Research\n\n${dedupedSections.join('\n\n---\n\n')}`;
+            }
+            if (navigationLinks.length > 0) {
+                const links = navigationLinks.map((l) => `- \`${l.code}\` opens ${l.url}`).join('\n');
+                result.text += `\n\n## Navigation Links\n\n${links}`;
+            }
+        }
+        async _discoverExpandables(researchText, missing = [], verifiedCodes = []) {
             const allElements = new Map();
             for (const section of parseResearchSections(researchText)) {
                 for (const el of section.elements) {
@@ -108,6 +171,15 @@ export function WithDeepAnalysis(Base) {
             if (allElements.size === 0)
                 return [];
             const eidxList = [...allElements.keys()].join(', ');
+            let missingHint = '';
+            if (missing.length > 0) {
+                const list = missing.map((s) => `- "${s.name}" (previously revealed via ${s.code})`).join('\n');
+                missingHint = dedent `
+
+          These sections were present on a previous visit but their trigger could not be replayed now — the element may have moved or been renamed. Prioritize finding the element that now reveals each:
+          ${list}
+        `;
+            }
             const textPrompt = dedent `
         From this UI research, identify elements that could reveal hidden UI when clicked
         (dropdown menus, popups, expandable panels, accordion sections, overflow menus, tab switches).
@@ -115,6 +187,7 @@ export function WithDeepAnalysis(Base) {
         Available eidx refs: ${eidxList}
 
         ${researchText}
+        ${missingHint}
 
         Rules:
         - Only pick elements that HIDE content until clicked (menus, dropdowns, accordions, tabs)
@@ -136,6 +209,7 @@ export function WithDeepAnalysis(Base) {
 
           Look for: overflow/ellipsis menus, chevron dropdowns, hamburger menus,
           gear/settings buttons, accordion toggles, tab switches, filter buttons.
+          ${missingHint}
 
           Rules:
           - For repeated icons (same icon on every list row), pick only the FIRST one
@@ -258,48 +332,15 @@ export function WithDeepAnalysis(Base) {
                             await this._restorePageState(state.url, originalAria);
                         }
                     }
-                    let clickCode = null;
-                    const action = this.explorer.createAction();
-                    for (const cmd of el.commands) {
-                        if (await action.attempt(cmd, undefined, false)) {
-                            clickCode = cmd;
-                            break;
-                        }
-                    }
-                    if (!clickCode) {
-                        debugLog(`Click failed: ${el.description.slice(0, 80)}`);
+                    const outcome = await this._executeAndAnalyze(el.commands, el.description, state, originalAria, this._summarizeExpanded(expandedSections));
+                    if (outcome.status === 'navigated') {
+                        navigationLinks.push({ code: outcome.code, url: outcome.url });
                         continue;
                     }
-                    await new Promise((r) => setTimeout(r, 500));
-                    let diff;
-                    try {
-                        await this.explorer.createAction().capturePageState();
-                        const currAR = ActionResult.fromState(this.stateManager.getCurrentState());
-                        diff = await currAR.diff(previousState);
-                        await diff.calculate();
-                    }
-                    catch (err) {
-                        tag('warning').log(`State capture failed after click: ${err instanceof Error ? err.message : err}`);
-                        await this._restorePageState(state.url, originalAria);
-                        continue;
-                    }
-                    if (diff.urlHasChanged()) {
-                        debugLog(`Click navigated to ${this.stateManager.getCurrentState()?.url}`);
-                        navigationLinks.push({ code: clickCode, url: this.stateManager.getCurrentState()?.url || '' });
-                        await this.navigateTo(state.url);
-                        continue;
-                    }
-                    const clickHtmlSize = diff.htmlParts.reduce((sum, p) => sum + p.subtree.length, 0);
-                    if (!diff.ariaChanged && clickHtmlSize <= 150) {
-                        debugLog(`No changes from: ${el.description.slice(0, 80)}`);
-                        continue;
-                    }
-                    const sectionMarkdown = await this._analyzeExpandedAction(clickCode, el.description, diff, this._summarizeExpanded(expandedSections));
-                    if (sectionMarkdown) {
-                        expandedSections.push(sectionMarkdown);
+                    if (outcome.status === 'revealed') {
+                        expandedSections.push(outcome.sectionMarkdown);
                         debugLog(`Captured section from: ${el.description.slice(0, 80)}`);
                     }
-                    await this._restorePageState(state.url, originalAria);
                 }
                 catch (err) {
                     tag('warning').log(`Expandable click failed for "${el.description.slice(0, 80)}": ${err instanceof Error ? err.message : err}`);
@@ -310,6 +351,50 @@ export function WithDeepAnalysis(Base) {
                 }
             }
         }
+        async _executeAndAnalyze(commands, description, state, originalAria, alreadyExpanded) {
+            const previousState = ActionResult.fromState(this.stateManager.getCurrentState());
+            let clickCode = null;
+            const action = this.explorer.createAction();
+            for (const cmd of commands) {
+                if (await action.attempt(cmd, undefined, false)) {
+                    clickCode = cmd;
+                    break;
+                }
+            }
+            if (!clickCode) {
+                debugLog(`Click failed: ${description.slice(0, 80)}`);
+                return { status: 'failed' };
+            }
+            await new Promise((r) => setTimeout(r, 500));
+            let diff;
+            try {
+                await this.explorer.createAction().capturePageState();
+                const currAR = ActionResult.fromState(this.stateManager.getCurrentState());
+                diff = await currAR.diff(previousState);
+                await diff.calculate();
+            }
+            catch (err) {
+                tag('warning').log(`State capture failed after click: ${err instanceof Error ? err.message : err}`);
+                await this._restorePageState(state.url, originalAria);
+                return { status: 'failed' };
+            }
+            if (diff.urlHasChanged()) {
+                const url = this.stateManager.getCurrentState()?.url || '';
+                debugLog(`Click navigated to ${url}`);
+                await this.navigateTo(state.url);
+                return { status: 'navigated', code: clickCode, url };
+            }
+            const clickHtmlSize = diff.htmlParts.reduce((sum, p) => sum + p.subtree.length, 0);
+            if (!diff.ariaChanged && clickHtmlSize <= 150) {
+                debugLog(`No changes from: ${description.slice(0, 80)}`);
+                return { status: 'none', code: clickCode };
+            }
+            const sectionMarkdown = await this._analyzeExpandedAction(clickCode, description, diff, alreadyExpanded);
+            await this._restorePageState(state.url, originalAria);
+            if (!sectionMarkdown)
+                return { status: 'none', code: clickCode };
+            return { status: 'revealed', code: clickCode, sectionMarkdown };
+        }
         async _restorePageState(url, originalAria) {
             try {
                 await this.cancelInUi();

package/dist/src/ai/researcher/locators.js

@@ -70,7 +70,7 @@ export function WithLocators(Base) {
                     broken++;
                 }
             }
-            tag('substep').log(`Validated ${locators.length} locators: ${locators.length - broken} valid, ${broken} broken`);
+            tag('operation').log(`Validated ${locators.length} locators: ${locators.length - broken} valid, ${broken} broken`);
         }
         async fixBrokenSections(result, conversation) {
             const broken = result.locators.filter((l) => l.valid === false);

package/dist/src/ai/researcher.js

@@ -114,7 +114,7 @@ export class Researcher extends ResearcherBase {
             if (!deep && !force) {
                 const similar = await findSimilarResearch(combinedHtml);
                 if (similar) {
-                    tag('substep').log('Similar research found, reusing cached result');
+                    tag('operation').log('Similar research found, reusing cached result');
                     if (stateHash)
                         saveResearch(stateHash, similar, combinedHtml);
                     tag('multiline').log(formatResearchSummary(similar));
@@ -264,10 +264,10 @@ export class Researcher extends ResearcherBase {
             tag('multiline').log(formatResearchSummary(result.text, { visionUsed: this.hasScreenshotToAnalyze }));
             tag('success').log('Research complete');
             if (researchFile)
-                tag('substep').log(`Research file saved to: ${researchFile}`);
+                tag('operation').log(`Research file saved to: ${researchFile}`);
             if (this.actionResult?.screenshotFile) {
                 const screenshotPath = outputPath('states', this.actionResult.screenshotFile);
-                tag('substep').log(`UI screenshot: file://${screenshotPath}`);
+                tag('operation').log(`UI screenshot: file://${screenshotPath}`);
             }
             await this.hooksRunner.runAfterHook('researcher', state.url);
             return result.text;
@@ -402,7 +402,7 @@ export class Researcher extends ResearcherBase {
                 .map((k) => k.content)
                 .filter((k) => !!k)
                 .join('\n\n');
-            tag('substep').log(`Found ${knowledgeFiles.length} relevant knowledge ${pluralize(knowledgeFiles.length, 'file')} for: ${this.actionResult.url}`);
+            tag('operation').log(`Found ${knowledgeFiles.length} relevant knowledge ${pluralize(knowledgeFiles.length, 'file')} for: ${this.actionResult.url}`);
             knowledge = `
         <hint>
         Here is relevant knowledge for this page:

package/dist/src/ai/session-analyst.js

@@ -34,6 +34,9 @@ export class SessionAnalyst {
 
       Crucial distinction: "the app misbehaved" vs "the automation could not interact with the app". ONLY the first is a Defect. If the automation gives up before the app responds — timeout, retries exhausted, dead loop / loop detected, could not click or find an element — that is an Execution issue regardless of what the log calls it. Failure inside the automation ≠ failure inside the product.
 
+      The action log is more authoritative than the scenario title. If the actual submitted data, page state, or action sequence does not match the scenario title, classify it as Execution issue and do not list that scenario under What works. Do NOT infer a product Defect or UX issue from behavior caused by incorrect test data or an automation mismatch.
+      Negative test data is valid when it matches a negative scenario. Do not call intentionally invalid input wrong data when the scenario expects rejection or validation feedback.
+
       A solitary failure where adjacent tests on the same feature passed → Execution, not Defect.
 
       ## Severity (defects only)
@@ -69,7 +72,7 @@ export class SessionAnalyst {
 
       ## Brevity rules
 
-      - Headline: 2 sentences MAX. About the FEATURE, not the run. No counts, no "N tests", no "this session". Banned words: "exercised", "comprehensive", "notably", "this session", "module", "targeted", "covered creation".
+      - Headline: 2 sentences MAX. About the FEATURE, not the run. No counts, no "N tests", no "this session". Never use these words: "exercised", "comprehensive", "notably", "this session", "module", "targeted", "covered creation".
       - What works: feature name + test refs. NO parentheticals, NO caveats. If there's a caveat, the entry doesn't belong here.
       - Defect title is the BUG ("Search returns non-matching results"), never the scenario name.
       - Reproduce steps are imperative one-liners drawn from the log.

package/dist/src/ai/task-agent.js

@@ -24,7 +24,7 @@ export class TaskAgent {
             .map((k) => k.content)
             .filter((k) => !!k)
             .join('\n\n');
-        tag('substep').log(`Found ${knowledgeFiles.length} relevant knowledge ${pluralize(knowledgeFiles.length, 'file')}`);
+        tag('operation').log(`Found ${knowledgeFiles.length} relevant knowledge ${pluralize(knowledgeFiles.length, 'file')}`);
         return dedent `
       <knowledge>
       Here is relevant knowledge for this page:
@@ -40,7 +40,7 @@ export class TaskAgent {
             return '';
         const totalSections = toc.reduce((sum, entry) => sum + entry.sections.length, 0);
         debugLog(`injecting experience TOC (${toc.length} files, ${totalSections} sections)`);
-        tag('substep').log(`Found ${toc.length} experience ${pluralize(toc.length, 'file')} (${totalSections} sections)`);
+        tag('operation').log(`Found ${toc.length} experience ${pluralize(toc.length, 'file')} (${totalSections} sections)`);
         return renderExperienceToc(toc);
     }
     setHistorian(historian) {

package/dist/src/commands/context-aria-command.js

@@ -12,6 +12,6 @@ export class ContextAriaCommand extends BaseCommand {
         if (!ariaSnapshot) {
             throw new Error('No ARIA snapshot available for current page');
         }
-        tag('multiline').log(`ARIA Snapshot:\n\n${ariaSnapshot}`);
+        tag('multiline').log(`ARIA Snapshot:\n\n${ariaSnapshot}`, { maxLines: 10 });
     }
 }