@massu/core 1.10.8 → 1.12.0

package/src/memory-file-ingest.ts

@@ -1,128 +0,0 @@
-// Copyright (c) 2026 Massu. All rights reserved.
-// Licensed under BSL 1.1 - see LICENSE file for details.
-
-// ============================================================
-// Memory File Auto-Ingest
-// Shared module for parsing memory/*.md files and ingesting
-// their YAML frontmatter + content into the observations table.
-// Used by: post-tool-use.ts, memory-tools.ts, init.ts
-// ============================================================
-
-import type Database from 'better-sqlite3';
-import { readFileSync, existsSync, readdirSync } from 'fs';
-import { join } from 'path';
-import { parse as parseYaml } from 'yaml';
-import { addObservation } from './memory-db.ts';
-
-export type IngestResult = 'inserted' | 'updated' | 'skipped';
-
-/**
- * Parse a memory/*.md file's YAML frontmatter and ingest it into the
- * observations table. Deduplicates by title prefix `[memory-file] {name}`.
- *
- * @returns 'inserted' | 'updated' | 'skipped'
- */
-export function ingestMemoryFile(
-  db: Database.Database,
-  sessionId: string,
-  filePath: string,
-): IngestResult {
-  if (!existsSync(filePath)) return 'skipped';
-
-  const content = readFileSync(filePath, 'utf-8');
-  const basename = (filePath.split('/').pop() ?? '').replace('.md', '');
-
-  // Parse YAML frontmatter (between first --- and second ---)
-  const frontmatterMatch = content.match(/^---\n([\s\S]*?)\n---/);
-
-  let name = basename;
-  let description = '';
-  let type = 'discovery';
-  let confidence: number | undefined;
-
-  if (frontmatterMatch) {
-    try {
-      // pattern-scanner-allow: yaml-parse — reason: parses YAML FRONTMATTER from markdown memory files (NOT massu.config.yaml). This is document metadata parsing, not application config access; getConfig() does not apply.
-      const fm = parseYaml(frontmatterMatch[1]) as Record<string, unknown>;
-      name = (fm.name as string) ?? basename;
-      description = (fm.description as string) ?? '';
-      type = (fm.type as string) ?? 'discovery';
-      confidence = fm.confidence != null ? Number(fm.confidence) : undefined;
-    } catch {
-      // Use defaults if YAML parsing fails
-    }
-  }
-
-  // Map memory types to observation types
-  const obsType = mapMemoryTypeToObservationType(type);
-
-  // Calculate importance from confidence (0.0-1.0 -> 1-5)
-  const importance = confidence != null
-    ? Math.max(1, Math.min(5, Math.round(confidence * 4 + 1)))
-    : 4;
-
-  // Extract body (after second ---)
-  const bodyMatch = content.match(/^---\n[\s\S]*?\n---\n([\s\S]*)/);
-  const body = bodyMatch ? bodyMatch[1].trim().slice(0, 500) : '';
-
-  const title = `[memory-file] ${name}`;
-  const detail = description ? `${description}\n\n${body}` : body;
-
-  // Deduplicate: check if this exact title exists
-  const existing = db.prepare(
-    'SELECT id FROM observations WHERE title = ? LIMIT 1'
-  ).get(title) as { id: number } | undefined;
-
-  if (existing) {
-    db.prepare('UPDATE observations SET detail = ?, importance = ? WHERE id = ?')
-      .run(detail, importance, existing.id);
-    return 'updated';
-  } else {
-    addObservation(db, sessionId, obsType, title, detail, { importance });
-    return 'inserted';
-  }
-}
-
-/**
- * Bulk-ingest all memory/*.md files from a directory.
- * Skips MEMORY.md (the index file).
- *
- * @returns { inserted, updated, skipped, total }
- */
-export function backfillMemoryFiles(
-  db: Database.Database,
-  memoryDir: string,
-  sessionId?: string,
-): { inserted: number; updated: number; skipped: number; total: number } {
-  const stats = { inserted: 0, updated: 0, skipped: 0, total: 0 };
-
-  if (!existsSync(memoryDir)) return stats;
-
-  const files = readdirSync(memoryDir).filter(
-    f => f.endsWith('.md') && f !== 'MEMORY.md'
-  );
-  stats.total = files.length;
-
-  const sid = sessionId ?? `backfill-${Date.now()}`;
-
-  for (const file of files) {
-    const result = ingestMemoryFile(db, sid, join(memoryDir, file));
-    stats[result]++;
-  }
-
-  return stats;
-}
-
-function mapMemoryTypeToObservationType(memoryType: string): string {
-  switch (memoryType) {
-    case 'user':
-    case 'feedback':
-      return 'decision';
-    case 'project':
-      return 'feature';
-    case 'reference':
-      return 'discovery';
-    default:
-      return 'discovery';
-  }
-}

package/src/memory-tools.ts

@@ -1,424 +0,0 @@
-// Copyright (c) 2026 Massu. All rights reserved.
-// Licensed under BSL 1.1 - see LICENSE file for details.
-
-import type Database from 'better-sqlite3';
-import type { ToolDefinition, ToolResult } from './tools.ts';
-import {
-  searchObservations,
-  getRecentObservations,
-  getSessionSummaries,
-  getSessionTimeline,
-  getFailedAttempts,
-  addObservation,
-  assignImportance,
-  createSession,
-} from './memory-db.ts';
-import { getConfig, getResolvedPaths } from './config.ts';
-import { backfillMemoryFiles } from './memory-file-ingest.ts';
-
-/** Prefix a base tool name with the configured tool prefix. */
-function p(baseName: string): string {
-  return `${getConfig().toolPrefix}_${baseName}`;
-}
-
-// ============================================================
-// P4-001 through P4-006: MCP Memory Tools
-// ============================================================
-
-/**
- * Get all memory tool definitions.
- */
-export function getMemoryToolDefinitions(): ToolDefinition[] {
-  return [
-    // P4-001: memory_search
-    {
-      name: p('memory_search'),
-      description: 'Search past session observations and decisions using full-text search. Returns compact index of matching observations.',
-      inputSchema: {
-        type: 'object',
-        properties: {
-          query: { type: 'string', description: 'Search text (FTS5 query syntax supported)' },
-          type: { type: 'string', description: 'Filter by observation type (decision, bugfix, feature, failed_attempt, cr_violation, vr_check, etc.)' },
-          cr_rule: { type: 'string', description: 'Filter by CR rule (e.g., CR-9)' },
-          date_from: { type: 'string', description: 'Start date (ISO format)' },
-          limit: { type: 'number', description: 'Max results (default: 20)' },
-        },
-        required: ['query'],
-      },
-    },
-    // P4-002: memory_timeline
-    {
-      name: p('memory_timeline'),
-      description: 'Retrieve episodic memory - chronological context around a specific event. Shows observations before and after the anchor point to reconstruct what happened and why.',
-      inputSchema: {
-        type: 'object',
-        properties: {
-          observation_id: { type: 'number', description: 'Anchor observation ID' },
-          depth_before: { type: 'number', description: 'Items before (default: 5)' },
-          depth_after: { type: 'number', description: 'Items after (default: 5)' },
-        },
-        required: ['observation_id'],
-      },
-    },
-    // P4-003: memory_detail
-    {
-      name: p('memory_detail'),
-      description: 'Get full observation details by IDs (batch). Includes evidence, files, plan items.',
-      inputSchema: {
-        type: 'object',
-        properties: {
-          ids: {
-            type: 'array',
-            items: { type: 'number' },
-            description: 'Observation IDs to retrieve',
-          },
-        },
-        required: ['ids'],
-      },
-    },
-    // P4-004: memory_sessions
-    {
-      name: p('memory_sessions'),
-      description: 'List recent sessions with summaries.',
-      inputSchema: {
-        type: 'object',
-        properties: {
-          limit: { type: 'number', description: 'Max sessions (default: 10)' },
-          status: { type: 'string', description: 'Filter by status (active, completed, abandoned)' },
-        },
-        required: [],
-      },
-    },
-    // P4-005: memory_failures
-    {
-      name: p('memory_failures'),
-      description: 'Get all failed attempts (DON\'T RETRY warnings). Check before attempting a fix to see if it was already tried.',
-      inputSchema: {
-        type: 'object',
-        properties: {
-          query: { type: 'string', description: 'Filter by keyword' },
-          limit: { type: 'number', description: 'Max results (default: 20)' },
-        },
-        required: [],
-      },
-    },
-    // P4-007: memory_backfill
-    {
-      name: p('memory_backfill'),
-      description: 'Scan all memory/*.md files and ingest into database. Run after massu init or to recover from DB loss. Parses YAML frontmatter and deduplicates by title.',
-      inputSchema: {
-        type: 'object',
-        properties: {},
-        required: [],
-      },
-    },
-    // P4-006: memory_ingest
-    {
-      name: p('memory_ingest'),
-      description: 'Manually record an observation that hooks cannot auto-detect. Use for significant decisions, discoveries, or failed attempts mid-session.',
-      inputSchema: {
-        type: 'object',
-        properties: {
-          type: {
-            type: 'string',
-            description: 'Observation type: decision, bugfix, feature, refactor, discovery, cr_violation, vr_check, pattern_compliance, failed_attempt, file_change, incident_near_miss',
-          },
-          title: { type: 'string', description: 'Short description' },
-          detail: { type: 'string', description: 'Full context' },
-          importance: { type: 'number', description: 'Override importance (1-5, default: auto-assigned)' },
-          cr_rule: { type: 'string', description: 'Link to CR rule (e.g., CR-9)' },
-          plan_item: { type: 'string', description: 'Link to plan item (e.g., P2-003)' },
-          files: {
-            type: 'array',
-            items: { type: 'string' },
-            description: 'Files involved',
-          },
-        },
-        required: ['type', 'title'],
-      },
-    },
-  ];
-}
-
-/**
- * Handle a memory tool call.
- */
-export function handleMemoryToolCall(
-  name: string,
-  args: Record<string, unknown>,
-  memoryDb: Database.Database
-): ToolResult {
-  try {
-    const prefix = getConfig().toolPrefix + '_';
-    const baseName = name.startsWith(prefix) ? name.slice(prefix.length) : name;
-
-    switch (baseName) {
-      case 'memory_search':
-        return handleSearch(args, memoryDb);
-      case 'memory_timeline':
-        return handleTimeline(args, memoryDb);
-      case 'memory_detail':
-        return handleDetail(args, memoryDb);
-      case 'memory_sessions':
-        return handleSessions(args, memoryDb);
-      case 'memory_failures':
-        return handleFailures(args, memoryDb);
-      case 'memory_ingest':
-        return handleIngest(args, memoryDb);
-      case 'memory_backfill':
-        return handleBackfill(memoryDb);
-      default:
-        return text(`Unknown memory tool: ${name}`);
-    }
-  } catch (error) {
-    return text(`Error in ${name}: ${error instanceof Error ? error.message : String(error)}`);
-  }
-}
-
-// ============================================================
-// Tool Handlers
-// ============================================================
-
-function handleSearch(args: Record<string, unknown>, db: Database.Database): ToolResult {
-  const query = args.query as string;
-  if (!query) return text('Error: query is required');
-
-  const results = searchObservations(db, query, {
-    type: args.type as string | undefined,
-    crRule: args.cr_rule as string | undefined,
-    dateFrom: args.date_from as string | undefined,
-    limit: args.limit as number | undefined,
-  });
-
-  if (results.length === 0) {
-    return text(`No observations found for "${query}".`);
-  }
-
-  const lines = [`## Search Results for "${query}" (${results.length} matches)`, ''];
-  lines.push('| ID | Type | Title | Date | Importance |');
-  lines.push('|----|------|-------|------|------------|');
-
-  for (const r of results) {
-    lines.push(`| ${r.id} | ${r.type} | ${r.title.slice(0, 80)} | ${r.created_at.split('T')[0]} | ${r.importance} |`);
-  }
-
-  lines.push('');
-  lines.push(`Use ${p('memory_detail')} with IDs for full details.`);
-  lines.push(`Use ${p('memory_timeline')} with an ID for chronological context.`);
-
-  return text(lines.join('\n'));
-}
-
-function handleTimeline(args: Record<string, unknown>, db: Database.Database): ToolResult {
-  const observationId = args.observation_id as number;
-  if (!observationId) return text('Error: observation_id is required');
-
-  const depthBefore = (args.depth_before as number) ?? 5;
-  const depthAfter = (args.depth_after as number) ?? 5;
-
-  // Get the anchor observation
-  const anchor = db.prepare('SELECT * FROM observations WHERE id = ?').get(observationId) as Record<string, unknown> | undefined;
-  if (!anchor) return text(`Observation ${observationId} not found.`);
-
-  // Get observations before
-  const before = db.prepare(
-    'SELECT id, type, title, created_at, importance FROM observations WHERE session_id = ? AND created_at_epoch < ? ORDER BY created_at_epoch DESC LIMIT ?'
-  ).all(anchor.session_id, anchor.created_at_epoch, depthBefore) as Array<Record<string, unknown>>;
-
-  // Get observations after
-  const after = db.prepare(
-    'SELECT id, type, title, created_at, importance FROM observations WHERE session_id = ? AND created_at_epoch > ? ORDER BY created_at_epoch ASC LIMIT ?'
-  ).all(anchor.session_id, anchor.created_at_epoch, depthAfter) as Array<Record<string, unknown>>;
-
-  const lines = [`## Timeline around observation #${observationId}`, ''];
-
-  // Before (reversed to chronological order)
-  for (const o of before.reverse()) {
-    lines.push(`  ${o.id} | ${o.type} | ${(o.title as string).slice(0, 60)} | ${(o.created_at as string).split('T')[0]}`);
-  }
-
-  // Anchor
-  lines.push(`> ${anchor.id} | ${anchor.type} | ${(anchor.title as string).slice(0, 60)} | ${(anchor.created_at as string).split('T')[0]} <-- ANCHOR`);
-
-  // After
-  for (const o of after) {
-    lines.push(`  ${o.id} | ${o.type} | ${(o.title as string).slice(0, 60)} | ${(o.created_at as string).split('T')[0]}`);
-  }
-
-  return text(lines.join('\n'));
-}
-
-function handleDetail(args: Record<string, unknown>, db: Database.Database): ToolResult {
-  const ids = args.ids as number[];
-  if (!ids || ids.length === 0) return text('Error: ids array is required');
-
-  const placeholders = ids.map(() => '?').join(',');
-  const observations = db.prepare(
-    `SELECT * FROM observations WHERE id IN (${placeholders}) ORDER BY created_at_epoch ASC`
-  ).all(...ids) as Array<Record<string, unknown>>;
-
-  if (observations.length === 0) {
-    return text('No observations found for the given IDs.');
-  }
-
-  const lines: string[] = [];
-  for (const o of observations) {
-    lines.push(`## Observation #${o.id} [${o.type}] (importance: ${o.importance})`);
-    lines.push(`**Title**: ${o.title}`);
-    lines.push(`**Session**: ${o.session_id}`);
-    lines.push(`**Date**: ${o.created_at}`);
-
-    if (o.detail) lines.push(`**Detail**: ${o.detail}`);
-    if (o.files_involved && o.files_involved !== '[]') {
-      const files = safeParseJson(o.files_involved as string, []) as string[];
-      if (files.length > 0) lines.push(`**Files**: ${files.join(', ')}`);
-    }
-    if (o.plan_item) lines.push(`**Plan Item**: ${o.plan_item}`);
-    if (o.cr_rule) lines.push(`**CR Rule**: ${o.cr_rule}`);
-    if (o.vr_type) lines.push(`**VR Type**: ${o.vr_type}`);
-    if (o.evidence) lines.push(`**Evidence**: ${(o.evidence as string).slice(0, 500)}`);
-    if ((o.recurrence_count as number) > 1) lines.push(`**Recurrence**: ${o.recurrence_count}x`);
-    lines.push('');
-  }
-
-  return text(lines.join('\n'));
-}
-
-function handleSessions(args: Record<string, unknown>, db: Database.Database): ToolResult {
-  const limit = (args.limit as number) ?? 10;
-  const status = args.status as string | undefined;
-
-  let sql = 'SELECT * FROM sessions';
-  const params: (string | number)[] = [];
-
-  if (status) {
-    sql += ' WHERE status = ?';
-    params.push(status);
-  }
-
-  sql += ' ORDER BY started_at_epoch DESC LIMIT ?';
-  params.push(limit);
-
-  const sessions = db.prepare(sql).all(...params) as Array<Record<string, unknown>>;
-
-  if (sessions.length === 0) {
-    return text('No sessions found.');
-  }
-
-  const lines = ['## Recent Sessions', ''];
-  lines.push('| Session ID | Status | Branch | Started | Plan |');
-  lines.push('|------------|--------|--------|---------|------|');
-
-  for (const s of sessions) {
-    const started = (s.started_at as string).split('T')[0];
-    const plan = s.plan_file ? (s.plan_file as string).split('/').pop() : '-';
-    lines.push(`| ${(s.session_id as string).slice(0, 8)}... | ${s.status} | ${s.git_branch ?? '-'} | ${started} | ${plan} |`);
-  }
-
-  // Add summaries for each session
-  for (const s of sessions) {
-    const summaries = getSessionSummaries(db, 1);
-    const summary = summaries.find(sm => sm.session_id === s.session_id);
-    if (summary) {
-      lines.push('');
-      lines.push(`### ${(s.session_id as string).slice(0, 8)}...`);
-      if (summary.request) lines.push(`**Task**: ${summary.request.slice(0, 200)}`);
-      if (summary.completed) lines.push(`**Completed**: ${summary.completed.slice(0, 200)}`);
-    }
-  }
-
-  return text(lines.join('\n'));
-}
-
-function handleFailures(args: Record<string, unknown>, db: Database.Database): ToolResult {
-  const query = args.query as string | undefined;
-  const limit = (args.limit as number) ?? 20;
-
-  const failures = getFailedAttempts(db, query, limit);
-
-  if (failures.length === 0) {
-    return text(query ? `No failed attempts found for "${query}".` : 'No failed attempts recorded.');
-  }
-
-  const lines = ['## Failed Attempts (DO NOT RETRY)', ''];
-
-  for (const f of failures) {
-    const recurrence = f.recurrence_count > 1 ? ` (occurred ${f.recurrence_count}x across sessions)` : '';
-    lines.push(`### #${f.id}: ${f.title}${recurrence}`);
-    if (f.detail) lines.push(f.detail.slice(0, 500));
-    lines.push(`Session: ${f.session_id.slice(0, 8)}... | Date: ${f.created_at.split('T')[0]}`);
-    lines.push('');
-  }
-
-  return text(lines.join('\n'));
-}
-
-function handleIngest(args: Record<string, unknown>, db: Database.Database): ToolResult {
-  const type = args.type as string;
-  const title = args.title as string;
-
-  if (!type || !title) return text('Error: type and title are required');
-
-  const validTypes = ['decision', 'bugfix', 'feature', 'refactor', 'discovery',
-    'cr_violation', 'vr_check', 'pattern_compliance', 'failed_attempt',
-    'file_change', 'incident_near_miss'];
-
-  if (!validTypes.includes(type)) {
-    return text(`Error: invalid type "${type}". Valid types: ${validTypes.join(', ')}`);
-  }
-
-  // We need a session_id - get the most recent active session
-  const activeSession = db.prepare(
-    "SELECT session_id FROM sessions WHERE status = 'active' ORDER BY started_at_epoch DESC LIMIT 1"
-  ).get() as { session_id: string } | undefined;
-
-  if (!activeSession) {
-    return text('Error: no active session found. Start a session first.');
-  }
-
-  const importance = (args.importance as number) ?? assignImportance(type);
-  const id = addObservation(db, activeSession.session_id, type, title, (args.detail as string) ?? null, {
-    importance,
-    crRule: args.cr_rule as string | undefined,
-    planItem: args.plan_item as string | undefined,
-    filesInvolved: args.files as string[] | undefined,
-  });
-
-  return text(`Observation #${id} recorded successfully.\nType: ${type}\nTitle: ${title}\nImportance: ${importance}\nSession: ${activeSession.session_id.slice(0, 8)}...`);
-}
-
-function handleBackfill(db: Database.Database): ToolResult {
-  const memoryDir = getResolvedPaths().memoryDir;
-  const stats = backfillMemoryFiles(db, memoryDir);
-
-  const lines = [
-    '## Memory Backfill Results',
-    '',
-    `- **Total files scanned**: ${stats.total}`,
-    `- **Inserted (new)**: ${stats.inserted}`,
-    `- **Updated (existing)**: ${stats.updated}`,
-    `- **Skipped (not found)**: ${stats.skipped}`,
-    '',
-    stats.total === 0
-      ? 'No memory files found in memory directory.'
-      : `Successfully processed ${stats.inserted + stats.updated} of ${stats.total} memory files.`,
-  ];
-
-  return text(lines.join('\n'));
-}
-
-// ============================================================
-// Helpers
-// ============================================================
-
-function text(content: string): ToolResult {
-  return { content: [{ type: 'text', text: content }] };
-}
-
-function safeParseJson(json: string, fallback: unknown): unknown {
-  try {
-    return JSON.parse(json);
-  } catch (_e) {
-    return fallback;
-  }
-}

package/src/middleware-tree.ts

@@ -1,71 +0,0 @@
-// Copyright (c) 2026 Massu. All rights reserved.
-// Licensed under BSL 1.1 - see LICENSE file for details.
-
-import type Database from 'better-sqlite3';
-import { getConfig } from './config.ts';
-import { existsSync } from 'fs';
-import { resolve } from 'path';
-import { t } from './lib/sql-table-names.ts';
-
-/**
- * Build the middleware import tree by tracing all transitive imports
- * from the middleware entry point. Any file in this tree is subject to
- * Edge Runtime restrictions (no Node.js deps).
- */
-export function buildMiddlewareTree(dataDb: Database.Database): number {
-  // Clear existing data
-  dataDb.exec(`DELETE FROM ${t('middleware_tree')}`);
-
-  const config = getConfig();
-  const middlewareFile = config.paths.middleware ?? 'src/middleware.ts';
-
-  // Only build if the middleware file path is configured and exists conceptually
-  if (!middlewareFile) return 0;
-
-  // BFS through import edges starting from middleware.ts
-  const visited = new Set<string>();
-  const queue: string[] = [middlewareFile];
-  visited.add(middlewareFile);
-
-  while (queue.length > 0) {
-    const current = queue.shift()!;
-
-    const imports = dataDb.prepare(
-      `SELECT target_file FROM ${t('imports')} WHERE source_file = ?`
-    ).all(current) as { target_file: string }[];
-
-    for (const imp of imports) {
-      if (!visited.has(imp.target_file) && imp.target_file.startsWith('src/')) {
-        visited.add(imp.target_file);
-        queue.push(imp.target_file);
-      }
-    }
-  }
-
-  // Store the tree
-  const insertStmt = dataDb.prepare(`INSERT INTO ${t('middleware_tree')} (file) VALUES (?)`);
-  const insertAll = dataDb.transaction(() => {
-    for (const file of visited) {
-      insertStmt.run(file);
-    }
-  });
-  insertAll();
-
-  return visited.size;
-}
-
-/**
- * Check if a file is in the middleware import tree.
- */
-export function isInMiddlewareTree(dataDb: Database.Database, file: string): boolean {
-  const result = dataDb.prepare(`SELECT 1 FROM ${t('middleware_tree')} WHERE file = ?`).get(file);
-  return result !== undefined;
-}
-
-/**
- * Get all files in the middleware import tree.
- */
-export function getMiddlewareTree(dataDb: Database.Database): string[] {
-  const rows = dataDb.prepare(`SELECT file FROM ${t('middleware_tree')} ORDER BY file`).all() as { file: string }[];
-  return rows.map(r => r.file);
-}