thumbgate 0.9.10

package/scripts/lesson-inference.js

@@ -0,0 +1,315 @@
+#!/usr/bin/env node
+'use strict';
+
+/**
+ * Lesson Inference — surrounding message context extraction + lesson linking.
+ *
+ * When a user gives thumbs up/down, this module:
+ * 1. Reads the surrounding conversation context (prior + following messages)
+ * 2. Infers what the lesson is from that context
+ * 3. Creates a structured lesson with a stable link
+ * 4. Provides data for the statusbar to show the most recent lesson
+ *
+ * Competing with Mem0: our advantage is local-first + structured inference,
+ * not just raw storage.
+ */
+
+const fs = require('fs');
+const path = require('path');
+const { resolveFeedbackDir } = require('./feedback-paths');
+const {
+  buildStableId,
+  extractFilePaths,
+  extractToolCalls,
+  extractErrors,
+} = require('./conversation-context');
+
+const LESSONS_FILE = 'lessons-index.jsonl';
+const RECENT_LESSON_FILE = 'recent-lesson.json';
+
+function getFeedbackDir() {
+  return resolveFeedbackDir();
+}
+
+function getLessonBaseUrl() {
+  const configuredOrigin = String(process.env.THUMBGATE_PUBLIC_APP_ORIGIN || '').trim().replace(/\/+$/, '');
+  return configuredOrigin || 'http://localhost:3456';
+}
+
+function getLessonsPath() { return path.join(getFeedbackDir(), LESSONS_FILE); }
+function getRecentLessonPath() { return path.join(getFeedbackDir(), RECENT_LESSON_FILE); }
+
+function readJsonl(fp) {
+  if (!fs.existsSync(fp)) return [];
+  const raw = fs.readFileSync(fp, 'utf-8').trim();
+  if (!raw) return [];
+  return raw.split('\n').map((l) => { try { return JSON.parse(l); } catch { return null; } }).filter(Boolean);
+}
+
+function ensureDir(p) { const d = path.dirname(p); if (!fs.existsSync(d)) fs.mkdirSync(d, { recursive: true }); }
+
+// ---------------------------------------------------------------------------
+// 1. Surrounding Message Context Extraction
+// ---------------------------------------------------------------------------
+
+/**
+ * Extract lesson context from surrounding messages.
+ * Takes the conversation turns before/after the feedback signal.
+ *
+ * @param {Object} opts
+ * @param {Array} opts.priorMessages - Messages before the feedback (most recent first)
+ * @param {Array} opts.followingMessages - Messages after the feedback
+ * @param {string} opts.signal - 'positive' or 'negative'
+ * @param {string} opts.feedbackContext - User-provided context string
+ * @returns {{ inferredLesson, triggerMessage, priorSummary, confidence }}
+ */
+function inferFromSurroundingMessages({ priorMessages = [], followingMessages = [], signal, feedbackContext = '' } = {}) {
+  const prior = priorMessages.slice(0, 5); // Last 5 messages before feedback
+  const following = followingMessages.slice(0, 3); // Next 3 messages after
+
+  // The trigger message is typically the last assistant message before feedback
+  const triggerMessage = prior.find((m) => m.role === 'assistant') || prior[0] || null;
+  const triggerText = triggerMessage ? (triggerMessage.content || triggerMessage.text || '') : '';
+
+  // Extract what the agent did
+  const actionPatterns = [
+    { regex: /(?:edited|modified|changed|updated)\s+(.+?)(?:\.|$)/i, type: 'edit' },
+    { regex: /(?:created|wrote|added)\s+(.+?)(?:\.|$)/i, type: 'create' },
+    { regex: /(?:ran|executed|running)\s+(.+?)(?:\.|$)/i, type: 'command' },
+    { regex: /(?:fixed|resolved|patched)\s+(.+?)(?:\.|$)/i, type: 'fix' },
+    { regex: /(?:deployed|pushed|merged)\s+(.+?)(?:\.|$)/i, type: 'deploy' },
+    { regex: /(?:deleted|removed|dropped)\s+(.+?)(?:\.|$)/i, type: 'delete' },
+  ];
+
+  let inferredAction = null;
+  for (const ap of actionPatterns) {
+    const match = triggerText.match(ap.regex);
+    if (match) { inferredAction = { type: ap.type, target: match[1].trim().slice(0, 100) }; break; }
+  }
+
+  // Build the lesson
+  const isNegative = signal === 'negative' || signal === 'down';
+  let inferredLesson;
+
+  if (isNegative && inferredAction) {
+    inferredLesson = `Avoid: ${inferredAction.type} on ${inferredAction.target}. ${feedbackContext || 'User signaled this approach failed.'}`;
+  } else if (!isNegative && inferredAction) {
+    inferredLesson = `Repeat: ${inferredAction.type} on ${inferredAction.target}. ${feedbackContext || 'User confirmed this approach works.'}`;
+  } else if (feedbackContext) {
+    inferredLesson = feedbackContext;
+  } else {
+    inferredLesson = `${isNegative ? 'Negative' : 'Positive'} signal on agent output. No specific action inferred.`;
+  }
+
+  // Summarize prior context
+  const priorSummary = prior.slice(0, 3).map((m) => {
+    const role = m.role || 'unknown';
+    const text = (m.content || m.text || '').slice(0, 80);
+    return `[${role}] ${text}`;
+  }).join(' → ');
+
+  // Confidence: higher if we have more context
+  const contextSignals = [
+    feedbackContext.length > 10,
+    !!inferredAction,
+    prior.length >= 2,
+    !!triggerMessage,
+  ];
+  const confidence = Math.round((contextSignals.filter(Boolean).length / contextSignals.length) * 100);
+
+  return { inferredLesson, triggerMessage: triggerText.slice(0, 200), priorSummary, inferredAction, confidence, signal };
+}
+
+// ---------------------------------------------------------------------------
+// 2. Lesson Index & Linking
+// ---------------------------------------------------------------------------
+
+/**
+ * Create a lesson record with a stable link and store in the index.
+ */
+function createLesson({ feedbackId, signal, inferredLesson, triggerMessage, priorSummary, confidence, tags = [], metadata = {} } = {}) {
+  const lesson = {
+    id: buildStableId('lesson'),
+    feedbackId: feedbackId || null,
+    signal: signal || 'unknown',
+    lesson: inferredLesson || '',
+    triggerMessage: triggerMessage || '',
+    priorSummary: priorSummary || '',
+    confidence: confidence || 0,
+    tags,
+    metadata,
+    createdAt: new Date().toISOString(),
+    link: null, // populated below
+  };
+
+  // Stable link: dashboard deep-link to this lesson
+  lesson.link = `${getLessonBaseUrl()}/lessons#${lesson.id}`;
+
+  const lessonsPath = getLessonsPath();
+  ensureDir(lessonsPath);
+  fs.appendFileSync(lessonsPath, JSON.stringify(lesson) + '\n');
+
+  // Update recent lesson for statusbar
+  const recentPath = getRecentLessonPath();
+  fs.writeFileSync(recentPath, JSON.stringify(lesson, null, 2) + '\n');
+
+  return lesson;
+}
+
+/**
+ * Get the most recent lesson (for statusbar display).
+ */
+function getRecentLesson() {
+  const p = getRecentLessonPath();
+  if (!fs.existsSync(p)) return null;
+  try { return JSON.parse(fs.readFileSync(p, 'utf-8')); } catch { return null; }
+}
+
+/**
+ * Search lessons by query text.
+ */
+function searchLessons({ query = '', limit = 10, signal } = {}) {
+  const lessons = readJsonl(getLessonsPath());
+  const tokens = query.toLowerCase().split(/\s+/).filter(Boolean);
+
+  return lessons
+    .filter((l) => !signal || l.signal === signal)
+    .map((l) => {
+      const haystack = `${l.lesson} ${l.triggerMessage} ${l.priorSummary} ${(l.tags || []).join(' ')}`.toLowerCase();
+      let score = 0;
+      for (const t of tokens) { if (t.length > 2 && haystack.includes(t)) score += 1; }
+      return { ...l, _score: score };
+    })
+    .filter((l) => tokens.length === 0 || l._score > 0)
+    .sort((a, b) => b._score - a._score || new Date(b.createdAt) - new Date(a.createdAt))
+    .slice(0, limit);
+}
+
+/**
+ * Get lesson stats.
+ */
+function getLessonStats() {
+  const lessons = readJsonl(getLessonsPath());
+  const positive = lessons.filter((l) => l.signal === 'positive' || l.signal === 'up').length;
+  const negative = lessons.filter((l) => l.signal === 'negative' || l.signal === 'down').length;
+  const avgConfidence = lessons.length > 0 ? Math.round(lessons.reduce((s, l) => s + (l.confidence || 0), 0) / lessons.length) : 0;
+  return { total: lessons.length, positive, negative, avgConfidence };
+}
+
+// ---------------------------------------------------------------------------
+// 3. Statusbar Data Provider
+// ---------------------------------------------------------------------------
+
+/**
+ * Get data for the Claude Code statusbar.
+ * Returns the most recent lesson with link, formatted for display.
+ */
+function getStatusbarLessonData() {
+  const recent = getRecentLesson();
+  if (!recent) return { hasLesson: false, text: null, link: null };
+
+  const emoji = (recent.signal === 'negative' || recent.signal === 'down') ? '👎' : '👍';
+  const truncated = recent.lesson.length > 60 ? recent.lesson.slice(0, 57) + '...' : recent.lesson;
+
+  return {
+    hasLesson: true,
+    text: `${emoji} ${truncated}`,
+    link: recent.link,
+    lessonId: recent.id,
+    confidence: recent.confidence,
+    createdAt: recent.createdAt,
+  };
+}
+
+// ---------------------------------------------------------------------------
+// 4. Structured IF/THEN Lesson Extraction (v0.9.4)
+// ---------------------------------------------------------------------------
+
+function inferStructuredLesson(conversationWindow, signal, context) {
+  const normalizedWindow = Array.isArray(conversationWindow) ? conversationWindow : [];
+  const userMessages = normalizedWindow.filter(m => m.role === 'user');
+  const assistantMessages = normalizedWindow.filter(m => m.role === 'assistant');
+  const lastUser = userMessages[userMessages.length - 1]?.content || '';
+  const lastAssistant = assistantMessages[assistantMessages.length - 1]?.content || '';
+  const filePaths = extractFilePaths(normalizedWindow);
+  const toolCalls = extractToolCalls(normalizedWindow);
+  const errorPatterns = extractErrors(normalizedWindow);
+
+  return {
+    format: 'if-then-v1',
+    trigger: extractTrigger(lastUser),
+    action: extractAction(lastAssistant, signal),
+    signal,
+    confidence: calculateConfidence(normalizedWindow, context),
+    scope: inferScope(filePaths, toolCalls),
+    examples: [{ userIntent: lastUser.slice(0, 300), assistantAction: lastAssistant.slice(0, 300), outcome: signal === 'positive' ? 'approved' : 'rejected' }],
+    metadata: { toolsUsed: toolCalls, filesInvolved: filePaths.slice(0, 10), errorPatterns: errorPatterns.slice(0, 5), conversationLength: normalizedWindow.length, inferredAt: new Date().toISOString() },
+  };
+}
+
+function extractTrigger(userMsg) {
+  const text = String(userMsg || '').trim();
+  const lower = text.toLowerCase();
+  const leadingPhrases = [
+    { phrases: ['fix ', 'debug ', 'solve ', 'investigate '], type: 'debugging' },
+    { phrases: ['implement ', 'add ', 'create ', 'build '], type: 'implementation' },
+    { phrases: ['why ', 'how ', 'what ', 'where '], type: 'question' },
+    { phrases: ['don\'t ', 'do not ', 'never ', 'stop ', 'avoid '], type: 'constraint' },
+  ];
+
+  for (const entry of leadingPhrases) {
+    const match = consumePhrase(lower, text, entry.phrases);
+    if (match) return { condition: match, type: entry.type };
+  }
+
+  const errorIndex = ['error', 'fail', 'crash', 'broken', 'wrong']
+    .map((token) => lower.indexOf(token))
+    .filter((index) => index >= 0)
+    .sort((a, b) => a - b)[0];
+  if (Number.isInteger(errorIndex)) {
+    return {
+      condition: text.slice(errorIndex).replace(/^[:\-\s]+/, '').slice(0, 120).trim() || text.slice(0, 120).trim(),
+      type: 'error-report',
+    };
+  }
+
+  return { condition: text.slice(0, 120).trim(), type: 'general' };
+}
+
+function extractAction(assistantMsg, signal) {
+  return signal === 'positive'
+    ? { type: 'do', description: `Repeat this approach: ${assistantMsg.slice(0, 200).trim()}` }
+    : { type: 'avoid', description: `Avoid this approach: ${assistantMsg.slice(0, 200).trim()}` };
+}
+
+function calculateConfidence(window, context) {
+  let s = 0.5;
+  if (window.length >= 3) s += 0.1;
+  if (window.length >= 5) s += 0.1;
+  if (context && context.length > 20) s += 0.1;
+  if (extractFilePaths(window).length > 0) s += 0.1;
+  return Math.min(s, 1.0);
+}
+
+function inferScope(filePaths, toolCalls) {
+  if (filePaths.length === 0 && toolCalls.length === 0) return 'global';
+  if (filePaths.length <= 2) return 'file-level';
+  return 'project-level';
+}
+
+function consumePhrase(lower, original, phrases) {
+  for (const phrase of phrases) {
+    if (!lower.startsWith(phrase)) continue;
+    const value = original.slice(phrase.length).replace(/^[:\-\s]+/, '').slice(0, 120).trim();
+    return value || original.slice(0, 120).trim();
+  }
+  return null;
+}
+
+module.exports = {
+  inferFromSurroundingMessages, createLesson, getRecentLesson,
+  searchLessons, getLessonStats, getStatusbarLessonData,
+  getLessonsPath, getRecentLessonPath,
+  inferStructuredLesson, extractTrigger, extractAction, extractToolCalls,
+  extractFilePaths, extractErrors, calculateConfidence, inferScope,
+};

package/scripts/lesson-retrieval.js

@@ -0,0 +1,95 @@
+#!/usr/bin/env node
+'use strict';
+
+/**
+ * Per-action lesson retrieval.
+ * Given a tool name + context, returns the top-K most relevant lessons
+ * using keyword matching + recency decay + signal weighting.
+ */
+
+const RECENCY_DECAY_DAYS = 30; // lessons older than this get down-weighted
+
+function retrieveRelevantLessons(toolName, actionContext, options = {}) {
+  const { maxResults = 5, feedbackDir } = options;
+  const { getFeedbackPaths, readJSONL } = require('./feedback-loop');
+  const pathMod = require('path');
+  const paths = feedbackDir
+    ? { MEMORY_LOG_PATH: pathMod.join(feedbackDir, 'memory-log.jsonl') }
+    : getFeedbackPaths();
+
+  const memories = readJSONL(paths.MEMORY_LOG_PATH, { maxLines: 200 });
+  if (memories.length === 0) return [];
+
+  // Score each memory against the current action
+  const scored = memories.map((mem) => ({
+    ...mem,
+    relevanceScore: scoreRelevance(mem, toolName, actionContext),
+  }));
+
+  // Sort by relevance, return top-K
+  return scored
+    .filter((m) => m.relevanceScore > 0.1)
+    .sort((a, b) => b.relevanceScore - a.relevanceScore)
+    .slice(0, maxResults)
+    .map((m) => ({
+      id: m.id,
+      title: m.title,
+      content: m.content,
+      signal: m.tags?.includes('negative') ? 'negative' : 'positive',
+      rule: m.structuredRule || null,
+      relevanceScore: m.relevanceScore,
+      timestamp: m.timestamp,
+    }));
+}
+
+function scoreRelevance(memory, toolName, actionContext) {
+  let score = 0;
+
+  const memText = `${memory.title || ''} ${memory.content || ''} ${(memory.tags || []).join(' ')}`.toLowerCase();
+  const contextLower = (actionContext || '').toLowerCase();
+  const toolLower = (toolName || '').toLowerCase();
+
+  // 1. Tool name match (high weight)
+  if (memory.metadata?.toolsUsed?.some((t) => t.toLowerCase() === toolLower)) score += 0.4;
+  if (memText.includes(toolLower)) score += 0.2;
+
+  // 2. File path overlap
+  const contextPaths = extractPaths(actionContext);
+  const memPaths = memory.metadata?.filesInvolved || extractPaths(memText);
+  const pathOverlap = contextPaths.filter((p) =>
+    memPaths.some((mp) => mp.includes(p) || p.includes(mp)),
+  );
+  if (pathOverlap.length > 0) score += 0.3;
+
+  // 3. Keyword overlap (TF-IDF-lite)
+  const contextTokens = tokenize(contextLower);
+  const memTokens = tokenize(memText);
+  const overlap = contextTokens.filter((t) => memTokens.includes(t));
+  score += Math.min(overlap.length * 0.05, 0.3);
+
+  // 4. Signal weighting — negative lessons are more important to surface
+  if (memory.tags?.includes('negative')) score += 0.1;
+
+  // 5. Recency decay
+  if (memory.timestamp) {
+    const ageMs = Date.now() - new Date(memory.timestamp).getTime();
+    const ageDays = ageMs / (1000 * 60 * 60 * 24);
+    const decay = Math.max(0, 1 - ageDays / RECENCY_DECAY_DAYS);
+    score *= 0.5 + 0.5 * decay; // 50% base + 50% recency
+  }
+
+  // 6. Structured rule bonus — IF/THEN rules are more actionable
+  if (memory.structuredRule) score += 0.15;
+
+  return score;
+}
+
+function extractPaths(text) {
+  return [...new Set((text || '').match(/(?:src\/|scripts\/|tests\/)[^\s,)'"<>]+/g) || [])];
+}
+
+function tokenize(text) {
+  return (text || '').split(/[\s.,;:!?()\[\]{}"'`]+/).filter((t) => t.length > 3);
+}
+
+module.exports = { retrieveRelevantLessons, scoreRelevance };

package/scripts/lesson-rotation.js

@@ -0,0 +1,137 @@
+'use strict';
+
+const STALE_THRESHOLD_DAYS = 60;
+const ARCHIVE_THRESHOLD_DAYS = 90;
+
+/**
+ * Add last_triggered and archived columns if they don't exist.
+ * Safe to call multiple times — uses IF NOT EXISTS logic via pragma.
+ */
+function migrateSchema(db) {
+  const columns = db.pragma('table_info(lessons)').map((c) => c.name);
+  if (!columns.includes('last_triggered')) {
+    db.exec('ALTER TABLE lessons ADD COLUMN last_triggered TEXT');
+  }
+  if (!columns.includes('archived')) {
+    db.exec('ALTER TABLE lessons ADD COLUMN archived INTEGER DEFAULT 0');
+  }
+  if (!columns.includes('trigger_count')) {
+    db.exec('ALTER TABLE lessons ADD COLUMN trigger_count INTEGER DEFAULT 0');
+  }
+}
+
+/**
+ * Record that a lesson was triggered (matched a gate or retrieved for context).
+ */
+function recordTrigger(db, lessonId) {
+  migrateSchema(db);
+  db.prepare(
+    'UPDATE lessons SET last_triggered = ?, trigger_count = COALESCE(trigger_count, 0) + 1 WHERE id = ?'
+  ).run(new Date().toISOString(), lessonId);
+}
+
+/**
+ * Score lesson staleness: 0 = fresh, 1 = completely stale.
+ */
+function stalenessScore(lesson) {
+  const ref = lesson.last_triggered || lesson.timestamp;
+  if (!ref) return 1;
+  const ageDays = (Date.now() - new Date(ref).getTime()) / (1000 * 60 * 60 * 24);
+  return Math.min(1, ageDays / ARCHIVE_THRESHOLD_DAYS);
+}
+
+/**
+ * Get all stale lessons (not triggered in STALE_THRESHOLD_DAYS).
+ */
+function findStaleLessons(db) {
+  migrateSchema(db);
+  const cutoff = new Date(Date.now() - STALE_THRESHOLD_DAYS * 86400000).toISOString();
+  return db.prepare(
+    `SELECT * FROM lessons
+     WHERE archived = 0 AND pruned = 0
+     AND (last_triggered IS NULL OR last_triggered < ?)
+     AND timestamp < ?
+     ORDER BY COALESCE(last_triggered, timestamp) ASC`
+  ).all(cutoff, cutoff);
+}
+
+/**
+ * Auto-archive lessons that haven't been triggered in ARCHIVE_THRESHOLD_DAYS.
+ * Returns { archived: number, reviewed: number }
+ */
+function autoArchive(db) {
+  migrateSchema(db);
+  const cutoff = new Date(Date.now() - ARCHIVE_THRESHOLD_DAYS * 86400000).toISOString();
+  const result = db.prepare(
+    `UPDATE lessons SET archived = 1
+     WHERE archived = 0 AND pruned = 0
+     AND (last_triggered IS NULL OR last_triggered < ?)
+     AND timestamp < ?`
+  ).run(cutoff, cutoff);
+  return { archived: result.changes };
+}
+
+/**
+ * Restore a lesson from archive.
+ */
+function restoreLesson(db, lessonId) {
+  migrateSchema(db);
+  db.prepare('UPDATE lessons SET archived = 0 WHERE id = ?').run(lessonId);
+}
+
+/**
+ * Get archived lessons for review.
+ */
+function getArchivedLessons(db) {
+  migrateSchema(db);
+  return db.prepare(
+    'SELECT * FROM lessons WHERE archived = 1 ORDER BY timestamp DESC'
+  ).all();
+}
+
+/**
+ * Generate a staleness report for the monthly review digest.
+ * Returns { stale: [...], archivable: [...], healthy: number }
+ */
+function stalenessReport(db) {
+  migrateSchema(db);
+  const staleThresholdDate = new Date(Date.now() - STALE_THRESHOLD_DAYS * 86400000).toISOString();
+  const archiveThresholdDate = new Date(Date.now() - ARCHIVE_THRESHOLD_DAYS * 86400000).toISOString();
+
+  const total = db.prepare('SELECT COUNT(*) as count FROM lessons WHERE archived = 0 AND pruned = 0').get().count;
+  const stale = findStaleLessons(db);
+  const archivable = stale.filter((l) => {
+    const ref = l.last_triggered || l.timestamp;
+    return ref && ref < archiveThresholdDate;
+  });
+
+  return {
+    total,
+    healthy: total - stale.length,
+    stale: stale.map((l) => ({
+      id: l.id,
+      context: (l.context || '').slice(0, 80),
+      importance: l.importance,
+      daysSinceActive: Math.round((Date.now() - new Date(l.last_triggered || l.timestamp).getTime()) / 86400000),
+      triggerCount: l.trigger_count || 0,
+    })),
+    archivable: archivable.map((l) => ({
+      id: l.id,
+      context: (l.context || '').slice(0, 80),
+      importance: l.importance,
+    })),
+  };
+}
+
+module.exports = {
+  migrateSchema,
+  recordTrigger,
+  stalenessScore,
+  findStaleLessons,
+  autoArchive,
+  restoreLesson,
+  getArchivedLessons,
+  stalenessReport,
+  STALE_THRESHOLD_DAYS,
+  ARCHIVE_THRESHOLD_DAYS,
+};