@mseep/core 3.0.0

package/lib/project-initializer.js

@@ -0,0 +1,856 @@
+import fs from 'fs';
+import os from 'os';
+import path from 'path';
+import https from 'https';
+import { fileURLToPath } from 'url';
+import { parseSkillFile } from './engine/parser.js';
+import { BookLibScanner } from './engine/scanner.js';
+import { resolveBookLibPaths } from './paths.js';
+import { loadConfig } from './config-loader.js';
+import { AgentDetector } from './agent-detector.js';
+
+const PACKAGE_ROOT = path.resolve(fileURLToPath(import.meta.url), '..', '..');
+
+const TOOL_FILE_MAP = {
+  claude:     { filePath: 'CLAUDE.md',                          fileHeader: '' },
+  cursor:     { filePath: '.cursor/rules/booklib-standards.mdc', fileHeader: null },
+  copilot:    { filePath: '.github/copilot-instructions.md',    fileHeader: '# Copilot Instructions\n\n' },
+  gemini:     { filePath: '.gemini/context.md',                 fileHeader: '# Project Context\n\n' },
+  codex:      { filePath: 'AGENTS.md',                          fileHeader: '# Agent Instructions\n\n' },
+  windsurf:   { filePath: '.windsurfrules',                     fileHeader: '' },
+  'roo-code': { filePath: '.roo/rules/booklib-standards.md',    fileHeader: null },
+  openhands:  { filePath: '.openhands/instructions.md',         fileHeader: '# OpenHands Instructions\n\n' },
+  junie:      { filePath: '.junie/guidelines.md',               fileHeader: '# Junie Guidelines\n\n' },
+  goose:      { filePath: '.goose/context.md',                  fileHeader: '# Goose Context\n\n' },
+  opencode:   { filePath: '.opencode/instructions.md',          fileHeader: '# OpenCode Instructions\n\n' },
+  letta:      { filePath: '.letta/skills/booklib.md',           fileHeader: null },
+};
+
+const TOOL_DOCS = {
+  claude:     'https://docs.anthropic.com/en/docs/claude-code/claude-md',
+  cursor:     'https://docs.cursor.com/context/rules-for-ai',
+  copilot:    'https://docs.github.com/en/copilot/customizing-copilot/adding-repository-custom-instructions',
+  gemini:     'https://github.com/google-gemini/gemini-cli#configuration',
+  codex:      'https://github.com/openai/codex#agents-md',
+  windsurf:   'https://docs.windsurf.com/windsurf/customize',
+  'roo-code': 'https://docs.roocode.com/features/custom-rules',
+  openhands:  'https://docs.all-hands.dev/usage/configuration',
+  junie:      'https://www.jetbrains.com/help/junie/guidelines',
+  goose:      'https://block.github.io/goose/docs/configuration',
+  opencode:   'https://github.com/opencode-ai/opencode#configuration',
+  letta:      'https://docs.letta.com/agents/custom-instructions',
+};
+
+/**
+ * Generates tool-specific context files from BookLib skills.
+ *
+ * Supported targets:
+ *   cursor   → .cursor/rules/booklib-standards.mdc
+ *   claude   → CLAUDE.md  (appends a standards section)
+ *   copilot  → .github/copilot-instructions.md
+ *   gemini   → .gemini/context.md
+ *   all      → all of the above
+ */
+
+const BOOKLIB_LINE = 'BookLib: knowledge tools for this project. Read skills/booklib-mcp-guide/SKILL.md before first use.';
+const AGENT_MARKER_START = '<!-- booklib-standards-start -->';
+const AGENT_MARKER_RE = /<!-- booklib-standards-start -->[\s\S]*?<!-- booklib-standards-end -->\n*/;
+
+const AGENT_SKELETON = `# Project
+
+## Stack
+<!-- describe your tech stack -->
+
+## Commands
+<!-- build, test, lint commands -->
+
+## Conventions
+<!-- coding standards, naming, patterns -->
+
+## Architecture
+<!-- key decisions and rationale -->
+
+`;
+
+export function writeAgentLine(filePath, opts = {}) {
+  const { skeleton = false } = opts;
+
+  if (!fs.existsSync(filePath)) {
+    fs.mkdirSync(path.dirname(filePath), { recursive: true });
+    const content = skeleton ? AGENT_SKELETON + BOOKLIB_LINE + '\n' : BOOKLIB_LINE + '\n';
+    fs.writeFileSync(filePath, content);
+    return 'created';
+  }
+
+  let content = fs.readFileSync(filePath, 'utf8');
+
+  // Already has the line — skip
+  if (content.includes('booklib-mcp-guide')) return 'skipped';
+
+  // Clean up old markers
+  if (content.includes(AGENT_MARKER_START)) {
+    content = content.replace(AGENT_MARKER_RE, '');
+  }
+
+  // Append the line
+  content = content.trimEnd() + '\n\n' + BOOKLIB_LINE + '\n';
+  fs.writeFileSync(filePath, content);
+  return 'updated';
+}
+
+export class ProjectInitializer {
+  constructor(options = {}) {
+    this.paths = resolveBookLibPaths(options.projectCwd);
+    this.projectCwd = options.projectCwd ?? process.cwd();
+    this.config = loadConfig(options.projectCwd);
+    this.scanner = new BookLibScanner();
+  }
+
+  /**
+   * Detects which skills are relevant to the project via scan, returns skill names.
+   */
+  detectRelevantSkills() {
+    const files = this.scanner.getFiles(this.projectCwd);
+    const seen = new Set();
+    for (const file of files) {
+      const skill = this.scanner.detectSkill(file);
+      if (skill) seen.add(skill);
+    }
+    return [...seen];
+  }
+
+  /**
+   * Main entry point. Detects or uses provided skills, then writes context files.
+   *
+   * @param {object} opts
+   * @param {string[]} [opts.skills]  - explicit skill names; auto-detected if omitted
+   * @param {string}   opts.target    - 'cursor' | 'claude' | 'copilot' | 'gemini' | 'all' | 'auto'
+   * @param {boolean}  [opts.dryRun]  - print what would be written, don't write
+   * @param {boolean}  [opts.quiet]   - suppress informational output (e.g. skipped files)
+   * @param {function} [opts.onFileConflict] - async callback invoked when a target file already
+   *   exists. Receives { filePath, lineCount, hasMarkers } and should return 'skip' to leave the
+   *   file untouched, or any other value ('append'|'update') to proceed with the default behavior.
+   *   When omitted, existing files are always appended/updated without prompting.
+   * @returns {string[]} list of files written
+   */
+  async init({ skills, target = 'all', dryRun = false, quiet = false, onFileConflict } = {}) {
+    const skillNames = skills?.length ? skills : this.detectRelevantSkills();
+    if (skillNames.length === 0) {
+      throw new Error('No relevant skills detected. Pass --skills explicitly or run booklib index first.');
+    }
+
+    const ALL_TARGETS = [
+      'claude', 'cursor', 'copilot', 'gemini', 'codex', 'windsurf',
+      'roo-code', 'openhands', 'junie', 'goose', 'opencode', 'letta',
+    ];
+    const targets = target === 'all'
+      ? ALL_TARGETS
+      : target === 'auto'
+      ? new AgentDetector({ cwd: this.projectCwd }).detect()
+      : target.split(',').map(t => t.trim());
+
+    const MARKER_START = '<!-- booklib-standards-start -->';
+    const MARKER_RE = /<!-- booklib-standards-start -->[\s\S]*?<!-- booklib-standards-end -->/;
+    const MARKER_END = '<!-- booklib-standards-end -->';
+
+    const blocks = this._extractBlocks(skillNames);
+
+    const written = [];
+    for (const t of targets) {
+      let filePath, content, fileHeader;
+
+      ({ filePath, content, fileHeader } = this._render(t, blocks, skillNames));
+
+      const absPath = path.join(this.projectCwd, filePath);
+      if (dryRun) {
+        console.log(`\n[dry-run] Would write: ${filePath}\n${'─'.repeat(60)}\n${content.slice(0, 400)}…`);
+      } else {
+        // Conflict detection: when file exists and a callback is provided, let the caller decide
+        if (fs.existsSync(absPath) && onFileConflict) {
+          const existing = fs.readFileSync(absPath, 'utf8');
+          const hasMarkers = existing.includes(MARKER_START);
+          const lineCount = existing.split('\n').length;
+          const action = await onFileConflict({ filePath, lineCount, hasMarkers });
+          if (action === 'skip') {
+            if (!quiet) console.log(`  · ${filePath} (skipped)`);
+            continue;
+          }
+          // 'append' or 'update' — fall through to existing write logic
+        }
+        fs.mkdirSync(path.dirname(absPath), { recursive: true });
+        if (fileHeader === null) {
+          // booklib owns this file entirely (cursor) — always overwrite
+          fs.writeFileSync(absPath, content);
+        } else if (fs.existsSync(absPath)) {
+          const existing = fs.readFileSync(absPath, 'utf8');
+          if (existing.includes(MARKER_START)) {
+            // Update only the booklib section, preserve everything else
+            fs.writeFileSync(absPath, existing.replace(MARKER_RE, content));
+          } else {
+            // File exists with no booklib section — append it
+            fs.appendFileSync(absPath, `\n\n${content}`);
+          }
+        } else {
+          // New file — write header + booklib section
+          fs.writeFileSync(absPath, `${fileHeader}${content}`);
+        }
+        written.push(filePath);
+        if (!quiet) console.log(`  ✅ ${filePath}`);
+      }
+    }
+    return written;
+  }
+
+  // ── Private helpers ────────────────────────────────────────────────────────
+
+  /**
+   * Reads each skill's SKILL.md and extracts framework + pitfall blocks.
+   */
+  _extractBlocks(skillNames) {
+    const blocks = [];
+    for (const name of skillNames) {
+      const skillPath    = path.join(this.paths.skillsPath, name, 'SKILL.md');
+      const cachePath    = path.join(this.paths.cachePath, 'skills', name, 'SKILL.md');
+      const bundledPath  = path.join(PACKAGE_ROOT, 'skills', name, 'SKILL.md');
+      const mdPath = [skillPath, cachePath, bundledPath].find(p => fs.existsSync(p)) ?? null;
+      if (!mdPath) continue;
+
+      const content = fs.readFileSync(mdPath, 'utf8');
+      const chunks = parseSkillFile(content, mdPath);
+
+      const framework = chunks.find(c =>
+        c.metadata.type === 'framework' || c.metadata.type === 'core_principles'
+      )?.text ?? null;
+
+      const pitfalls = chunks.find(c =>
+        c.metadata.type === 'pitfalls' || c.metadata.type === 'anti_patterns'
+      )?.text ?? null;
+
+      if (framework || pitfalls) {
+        blocks.push({ skill: name, framework, pitfalls });
+      }
+    }
+    return blocks;
+  }
+
+  /**
+   * Resolves the path to a skill's SKILL.md file by checking four locations
+   * in priority order: project-local, cached community, bundled, and Claude Code.
+   *
+   * @param {string} name - skill name (e.g. 'effective-kotlin')
+   * @returns {string|null} absolute path to SKILL.md, or null if not found
+   */
+  _findSkillFile(name) {
+    const candidates = [
+      path.join(this.paths.skillsPath, name, 'SKILL.md'),
+      path.join(this.paths.cachePath, 'skills', name, 'SKILL.md'),
+      path.join(PACKAGE_ROOT, 'skills', name, 'SKILL.md'),
+      path.join(os.homedir(), '.claude', 'skills', name, 'SKILL.md'),
+    ];
+    return candidates.find(p => fs.existsSync(p)) ?? null;
+  }
+
+  /**
+   * Renders extracted blocks into a tool-specific file.
+   */
+  _render(target, blocks, skillNames) {
+    const principles = blocks
+      .filter(b => b.framework)
+      .map(b => `### From ${b.skill}\n\n${b.framework.trim()}`)
+      .join('\n\n---\n\n');
+
+    const antiPatterns = blocks
+      .filter(b => b.pitfalls)
+      .map(b => `### From ${b.skill}\n\n${b.pitfalls.trim()}`)
+      .join('\n\n---\n\n');
+
+    const sources = skillNames.join(', ');
+    const generated = `Generated by BookLib from: ${sources}`;
+
+    const toolDocsUrl = TOOL_DOCS[target] ?? '';
+    const referencesSection = `
+### References
+
+- ${toolDocsUrl ? `[How to customize this file](${toolDocsUrl})` : 'Customize this file for your project'}
+- [BookLib documentation](https://booklib-ai.github.io/booklib/)
+- [BookLib skills catalog](https://github.com/booklib-ai/booklib)
+`;
+
+    switch (target) {
+      case 'cursor':
+        // booklib owns .cursor/rules/booklib-standards.mdc — full overwrite, fileHeader: null signals that
+        return {
+          filePath: '.cursor/rules/booklib-standards.mdc',
+          fileHeader: null,
+          content: `---
+description: Coding standards synthesized from ${sources}
+alwaysApply: true
+---
+
+<!-- ${generated} -->
+
+## Core Principles
+
+${principles || '_No structured principles found in selected skills._'}
+
+## Anti-Patterns to Avoid
+
+${antiPatterns || '_No anti-patterns found in selected skills._'}
+${referencesSection}`,
+        };
+
+      case 'claude':
+        return {
+          filePath: 'CLAUDE.md',
+          fileHeader: '',
+          content: `<!-- booklib-standards-start -->
+## Coding Standards
+
+> ${generated}
+
+### Principles
+
+${principles || '_No structured principles found._'}
+
+### Anti-Patterns
+
+${antiPatterns || '_No anti-patterns found._'}
+${referencesSection}<!-- booklib-standards-end -->`,
+        };
+
+      case 'copilot':
+        return {
+          filePath: '.github/copilot-instructions.md',
+          fileHeader: '# Copilot Instructions\n\n',
+          content: `<!-- booklib-standards-start -->
+<!-- ${generated} -->
+
+## What to follow
+
+${principles || '_No structured principles found._'}
+
+## What to avoid
+
+${antiPatterns || '_No anti-patterns found._'}
+${referencesSection}<!-- booklib-standards-end -->
+`,
+        };
+
+      case 'gemini':
+        return {
+          filePath: '.gemini/context.md',
+          fileHeader: '# Project Context\n\n',
+          content: `<!-- booklib-standards-start -->
+<!-- ${generated} -->
+
+## Coding Standards
+
+${principles || '_No structured principles found._'}
+
+## Anti-Patterns
+
+${antiPatterns || '_No anti-patterns found._'}
+${referencesSection}<!-- booklib-standards-end -->
+`,
+        };
+
+      case 'codex':
+        return {
+          filePath: 'AGENTS.md',
+          fileHeader: '# Agent Instructions\n\n',
+          content: `<!-- booklib-standards-start -->
+<!-- ${generated} -->
+
+## Coding Standards
+
+${principles || '_No structured principles found in selected skills._'}
+
+## Anti-Patterns to Avoid
+
+${antiPatterns || '_No anti-patterns found in selected skills._'}
+${referencesSection}<!-- booklib-standards-end -->
+`,
+        };
+
+      case 'windsurf':
+        return {
+          filePath: '.windsurfrules',
+          fileHeader: '',
+          content: `<!-- booklib-standards-start -->
+<!-- ${generated} -->
+
+## Core Principles
+
+${principles || '_No structured principles found in selected skills._'}
+
+## Anti-Patterns to Avoid
+
+${antiPatterns || '_No anti-patterns found in selected skills._'}
+${referencesSection}<!-- booklib-standards-end -->
+`,
+        };
+
+      case 'roo-code':
+        return {
+          filePath: '.roo/rules/booklib-standards.md',
+          fileHeader: null,
+          content: `<!-- booklib-standards-start -->
+<!-- ${generated} -->
+
+## Core Principles
+
+${principles || '_No structured principles found in selected skills._'}
+
+## Anti-Patterns
+
+${antiPatterns || '_No anti-patterns found._'}
+${referencesSection}<!-- booklib-standards-end -->
+`,
+        };
+
+      case 'openhands':
+        return {
+          filePath: '.openhands/instructions.md',
+          fileHeader: '# OpenHands Instructions\n\n',
+          content: `<!-- booklib-standards-start -->
+<!-- ${generated} -->
+
+## Coding Standards
+
+${principles || '_No structured principles found._'}
+
+## What to Avoid
+
+${antiPatterns || '_No anti-patterns found._'}
+${referencesSection}<!-- booklib-standards-end -->
+`,
+        };
+
+      case 'junie':
+        return {
+          filePath: '.junie/guidelines.md',
+          fileHeader: '# Junie Guidelines\n\n',
+          content: `<!-- booklib-standards-start -->
+<!-- ${generated} -->
+
+## Principles
+
+${principles || '_No structured principles found._'}
+
+## Anti-Patterns
+
+${antiPatterns || '_No anti-patterns found._'}
+${referencesSection}<!-- booklib-standards-end -->
+`,
+        };
+
+      case 'goose':
+        return {
+          filePath: '.goose/context.md',
+          fileHeader: '# Goose Context\n\n',
+          content: `<!-- booklib-standards-start -->
+<!-- ${generated} -->
+
+## Standards
+
+${principles || '_No structured principles found._'}
+
+## Avoid
+
+${antiPatterns || '_No anti-patterns found._'}
+${referencesSection}<!-- booklib-standards-end -->
+`,
+        };
+
+      case 'opencode':
+        return {
+          filePath: '.opencode/instructions.md',
+          fileHeader: '# OpenCode Instructions\n\n',
+          content: `<!-- booklib-standards-start -->
+<!-- ${generated} -->
+
+## Coding Standards
+
+${principles || '_No structured principles found._'}
+
+## Anti-Patterns
+
+${antiPatterns || '_No anti-patterns found._'}
+${referencesSection}<!-- booklib-standards-end -->
+`,
+        };
+
+      case 'letta':
+        return {
+          filePath: '.letta/skills/booklib.md',
+          fileHeader: null,
+          content: `<!-- booklib-standards-start -->
+<!-- ${generated} -->
+
+# BookLib Knowledge
+
+## Principles
+
+${principles || '_No structured principles found._'}
+
+## Anti-Patterns
+
+${antiPatterns || '_No anti-patterns found._'}
+${referencesSection}<!-- booklib-standards-end -->
+`,
+        };
+
+      default:
+        throw new Error(
+          `Unknown target: ${target}. Valid values: claude, cursor, copilot, gemini, codex, windsurf, roo-code, openhands, junie, goose, opencode, letta, all, auto`
+        );
+    }
+  }
+
+  /**
+   * Returns additional skill names worth knowing about, based on detected skills
+   * and project characteristics. Used for discovery hints — NOT injected into CLAUDE.md.
+   */
+  suggestRelatedSkills(detectedSkills, projectCwd) {
+    const suggestions = new Set();
+    const packageJsonPath = path.join(projectCwd, 'package.json');
+    let pkg = {};
+    try { pkg = JSON.parse(fs.readFileSync(packageJsonPath, 'utf8')); } catch { /* no package.json */ }
+    const deps = Object.keys({ ...pkg.dependencies, ...pkg.devDependencies });
+
+    for (const skill of detectedSkills) {
+      if (skill === 'clean-code-reviewer') {
+        suggestions.add('node-error-handling');
+        if (deps.some(d => ['express', 'fastify', 'koa', 'hono'].includes(d))) {
+          suggestions.add('owasp-input-validation');
+          suggestions.add('node-security-validation');
+        }
+      }
+      if (skill === 'effective-typescript') {
+        suggestions.add('clean-code-reviewer');
+        suggestions.add('node-error-handling');
+      }
+      if (skill === 'effective-python') {
+        suggestions.add('django-security');
+      }
+      if (skill === 'effective-java') {
+        suggestions.add('springboot-patterns');
+      }
+      if (skill === 'effective-kotlin') {
+        suggestions.add('kotlin-testing');
+      }
+    }
+
+    // Remove already-detected skills from suggestions
+    for (const s of detectedSkills) suggestions.delete(s);
+    return [...suggestions].slice(0, 3);
+  }
+
+  // ── ECC Artifact fetching ──────────────────────────────────────────────────
+
+  /**
+   * Pulls rules/, agents/, and commands/ from all configured github-skills-dir sources.
+   *
+   * Sources opt in by including an "artifacts" array in booklib.config.json:
+   *   { "type": "github-skills-dir", "repo": "...", "artifacts": ["rules", "agents", "commands"] }
+   *
+   * When multiple sources export the same artifact type, files are prefixed with the repo
+   * slug (last path segment of owner/repo) to avoid collisions.
+   *
+   * @param {object}   opts
+   * @param {string[]|null} [opts.languages]       - language folders for rules/ (null = all)
+   * @param {boolean}  [opts.includeAgents]        - pull agents/ → .claude/agents/
+   * @param {boolean}  [opts.includeCommands]      - pull commands/ → .claude/commands/
+   * @param {boolean}  [opts.dryRun]               - print what would be written without writing
+   * @returns {string[]} list of files written
+   */
+  async fetchEccArtifacts({ languages = null, includeAgents = true, includeCommands = true, dryRun = false } = {}) {
+    const artifactSources = this.config.sources.filter(
+      s => s.type === 'github-skills-dir' && Array.isArray(s.artifacts) && s.artifacts.length > 0
+    );
+
+    if (artifactSources.length === 0) {
+      throw new Error(
+        'No artifact-capable sources found. Add "artifacts": ["rules","agents","commands"] to a ' +
+        'github-skills-dir entry in booklib.config.json.'
+      );
+    }
+
+    // Determine which artifact types appear in more than one source — those need a prefix.
+    const typeCounts = { rules: 0, agents: 0, commands: 0 };
+    for (const src of artifactSources) {
+      if (src.artifacts.includes('rules'))    typeCounts.rules++;
+      if (src.artifacts.includes('agents'))   typeCounts.agents++;
+      if (src.artifacts.includes('commands')) typeCounts.commands++;
+    }
+
+    const written = [];
+
+    for (const source of artifactSources) {
+      const { repo, branch = 'main', artifacts: artifactList } = source;
+      // Derive a short slug from the repo name (owner/repo → repo segment)
+      const slug = repo.split('/').pop().replace(/[^a-z0-9]/gi, '-').toLowerCase();
+
+      if (languages !== false && artifactList.includes('rules')) {
+        const prefix = typeCounts.rules > 1 ? `${slug}-` : '';
+        written.push(...await this._pullRules(repo, branch, languages, dryRun, prefix));
+      }
+      if (includeAgents && artifactList.includes('agents')) {
+        const prefix = typeCounts.agents > 1 ? `${slug}-` : '';
+        written.push(...await this._pullDir(repo, branch, 'agents', '.claude/agents', dryRun, prefix));
+      }
+      if (includeCommands && artifactList.includes('commands')) {
+        const prefix = typeCounts.commands > 1 ? `${slug}-` : '';
+        written.push(...await this._pullDir(repo, branch, 'commands', '.claude/commands', dryRun, prefix));
+      }
+    }
+
+    return written;
+  }
+
+  /** Pulls rules/<language>/*.md → .cursor/rules/[prefix]<language>-<file>.mdc */
+  async _pullRules(repo, branch, languages, dryRun, prefix = '') {
+    const written = [];
+    let langDirs;
+    try {
+      const entries = await this._fetchJson(`https://api.github.com/repos/${repo}/contents/rules`);
+      if (!Array.isArray(entries)) return [];
+      langDirs = entries.filter(e => e.type === 'dir').map(e => e.name);
+    } catch {
+      return [];
+    }
+
+    if (languages && languages.length > 0) {
+      langDirs = langDirs.filter(d => languages.includes(d));
+    }
+
+    for (const lang of langDirs) {
+      let files;
+      try {
+        const entries = await this._fetchJson(`https://api.github.com/repos/${repo}/contents/rules/${lang}`);
+        files = Array.isArray(entries) ? entries.filter(e => e.type === 'file' && e.name.endsWith('.md')) : [];
+      } catch {
+        continue;
+      }
+
+      for (const file of files) {
+        const rawUrl = `https://raw.githubusercontent.com/${repo}/${branch}/rules/${lang}/${file.name}`;
+        const destName = `${prefix}${lang}-${file.name.replace(/\.md$/, '.mdc')}`;
+        const destPath = `.cursor/rules/${destName}`;
+        const absPath = path.join(this.projectCwd, destPath);
+
+        if (dryRun) {
+          console.log(`[dry-run] Would write: ${destPath} (from ${rawUrl})`);
+          written.push(destPath);
+          continue;
+        }
+
+        let content;
+        try { content = await this._fetchText(rawUrl); } catch { continue; }
+
+        if (!content.trimStart().startsWith('---')) {
+          content = `---\ndescription: ${lang} ${file.name.replace('.md', '')} rules\nalwaysApply: false\n---\n\n${content}`;
+        }
+
+        fs.mkdirSync(path.dirname(absPath), { recursive: true });
+        fs.writeFileSync(absPath, content);
+        console.log(`  ✅ ${destPath}`);
+        written.push(destPath);
+      }
+    }
+
+    return written;
+  }
+
+  /** Pulls <srcDir>/*.md → <destDir>/[prefix]<file>.md */
+  async _pullDir(repo, branch, srcDir, destDir, dryRun, prefix = '') {
+    const written = [];
+    let files;
+    try {
+      const entries = await this._fetchJson(`https://api.github.com/repos/${repo}/contents/${srcDir}`);
+      files = Array.isArray(entries) ? entries.filter(e => e.type === 'file' && e.name.endsWith('.md')) : [];
+    } catch {
+      return [];
+    }
+
+    for (const file of files) {
+      const rawUrl = `https://raw.githubusercontent.com/${repo}/${branch}/${srcDir}/${file.name}`;
+      const destPath = `${destDir}/${prefix}${file.name}`;
+      const absPath = path.join(this.projectCwd, destPath);
+
+      if (dryRun) {
+        console.log(`[dry-run] Would write: ${destPath} (from ${rawUrl})`);
+        written.push(destPath);
+        continue;
+      }
+
+      let content;
+      try { content = await this._fetchText(rawUrl); } catch { continue; }
+
+      fs.mkdirSync(path.dirname(absPath), { recursive: true });
+      fs.writeFileSync(absPath, content);
+      console.log(`  ✅ ${destPath}`);
+      written.push(destPath);
+    }
+
+    return written;
+  }
+
+  /**
+   * Writes MCP server config files for the selected tools.
+   *
+   * @param {object}   opts
+   * @param {string[]} opts.tools   - tool names: 'claude'|'cursor'|'copilot'|'gemini'|'codex'|'roo-code'|'windsurf'|'goose'|'zed'|'continue'
+   * @param {boolean}  [opts.dryRun]
+   * @returns {string[]} list of files written
+   */
+  async generateMcpConfigs({ tools = [], dryRun = false } = {}) {
+    const written = [];
+    for (const tool of tools) {
+      const config = this._renderMcpConfig(tool);
+      if (!config) continue;
+      const { filePath, mode } = config;
+      const absPath = config.global ? filePath : path.join(this.projectCwd, filePath);
+      if (dryRun) {
+        console.log(`[dry-run] Would write MCP config: ${filePath}`);
+        written.push(filePath);
+        continue;
+      }
+      fs.mkdirSync(path.dirname(absPath), { recursive: true });
+      if (mode === 'json-merge') {
+        this._mergeJsonMcpServer(absPath, filePath, config.mcpKey, config.mcpValue);
+      } else if (mode === 'toml-merge') {
+        this._mergeTomlMcpSection(absPath);
+      } else if (mode === 'yaml-merge') {
+        this._mergeGooseYaml(absPath);
+      } else {
+        fs.writeFileSync(absPath, config.content);
+      }
+      console.log(`  ✅ ${filePath}`);
+      written.push(filePath);
+    }
+    return written;
+  }
+
+  /** Returns a descriptor for writing the MCP config for a given tool. */
+  _renderMcpConfig(tool) {
+    const BOOKLIB_ENTRY = { command: 'booklib-mcp', args: [] };
+    switch (tool) {
+      case 'claude':
+        return { filePath: '.claude/settings.json',  mode: 'json-merge', mcpKey: ['mcpServers', 'booklib'],          mcpValue: BOOKLIB_ENTRY };
+      case 'cursor':
+        return { filePath: '.cursor/mcp.json',        mode: 'json-merge', mcpKey: ['mcpServers', 'booklib'],         mcpValue: BOOKLIB_ENTRY };
+      case 'copilot':
+        return { filePath: '.vscode/mcp.json',        mode: 'json-merge', mcpKey: ['servers', 'booklib'],            mcpValue: BOOKLIB_ENTRY };
+      case 'gemini':
+        return { filePath: '.gemini/settings.json',   mode: 'json-merge', mcpKey: ['mcpServers', 'booklib'],         mcpValue: BOOKLIB_ENTRY };
+      case 'codex':
+        return { filePath: '.codex/config.toml',      mode: 'toml-merge' };
+      case 'roo-code':
+        return { filePath: '.roo/mcp.json',            mode: 'json-merge', mcpKey: ['mcpServers', 'booklib'],        mcpValue: BOOKLIB_ENTRY };
+      case 'windsurf': {
+        const windsurfPath = path.join(os.homedir(), '.codeium', 'windsurf', 'mcp_config.json');
+        return { filePath: windsurfPath,              mode: 'json-merge', mcpKey: ['mcpServers', 'booklib'],         mcpValue: BOOKLIB_ENTRY, global: true };
+      }
+      case 'goose':
+        return { filePath: '.goose/config.yaml',      mode: 'yaml-merge' };
+      case 'zed':
+        return { filePath: '.zed/settings.json',      mode: 'json-merge', mcpKey: ['context_servers', 'booklib-mcp'], mcpValue: { command: { path: 'booklib-mcp', args: [] } } };
+      case 'continue':
+        return { filePath: '.continue/mcpServers/booklib.yaml', mode: 'overwrite', content: 'name: booklib\ncommand: booklib-mcp\nargs: []\n' };
+      default:
+        return null;
+    }
+  }
+
+  /** Reads an existing JSON config (if any), sets keyPath to value, writes back. */
+  _mergeJsonMcpServer(absPath, filePath, keyPath, value) {
+    let root = {};
+    if (fs.existsSync(absPath)) {
+      try {
+        root = JSON.parse(fs.readFileSync(absPath, 'utf8'));
+      } catch {
+        console.warn(`  ⚠️  Could not parse ${filePath} — writing fresh`);
+        root = {};
+      }
+    }
+    let node = root;
+    for (let i = 0; i < keyPath.length - 1; i++) {
+      if (!node[keyPath[i]] || typeof node[keyPath[i]] !== 'object') node[keyPath[i]] = {};
+      node = node[keyPath[i]];
+    }
+    node[keyPath[keyPath.length - 1]] = value;
+    fs.writeFileSync(absPath, JSON.stringify(root, null, 2) + '\n');
+  }
+
+  /** Appends or replaces the [mcp_servers.booklib] section in a TOML file. */
+  _mergeTomlMcpSection(absPath) {
+    const BOOKLIB_BLOCK = '[mcp_servers.booklib]\ncommand = "booklib-mcp"\nargs = []\n';
+    // Match from the section header to the next section header or end of string
+    const SECTION_RE = /\[mcp_servers\.booklib\][\s\S]*?(?=\n\[|$)/;
+
+    let existing = '';
+    if (fs.existsSync(absPath)) {
+      existing = fs.readFileSync(absPath, 'utf8');
+    }
+
+    if (SECTION_RE.test(existing)) {
+      fs.writeFileSync(absPath, existing.replace(SECTION_RE, BOOKLIB_BLOCK.trimEnd()));
+    } else {
+      fs.writeFileSync(absPath, existing + (existing.endsWith('\n') ? '' : '\n') + '\n' + BOOKLIB_BLOCK);
+    }
+  }
+
+  /** Appends or merges the booklib entry in a Goose YAML config. */
+  _mergeGooseYaml(absPath) {
+    const entry = '\nmcp_servers:\n  booklib:\n    command: booklib-mcp\n    args: []\n';
+    if (fs.existsSync(absPath)) {
+      const content = fs.readFileSync(absPath, 'utf8');
+      if (content.includes('booklib:')) return; // already exists
+      if (content.includes('mcp_servers:')) {
+        // Append under existing mcp_servers section
+        const updated = content.replace('mcp_servers:', 'mcp_servers:\n  booklib:\n    command: booklib-mcp\n    args: []');
+        fs.writeFileSync(absPath, updated);
+      } else {
+        fs.appendFileSync(absPath, entry);
+      }
+    } else {
+      fs.writeFileSync(absPath, entry.trim() + '\n');
+    }
+  }
+
+  // ── HTTP helpers ───────────────────────────────────────────────────────────
+
+  _fetchJson(url) {
+    return new Promise((resolve, reject) => {
+      https.get(url, { headers: { 'User-Agent': 'booklib-init/1.0' } }, res => {
+        if (res.statusCode >= 300 && res.statusCode < 400 && res.headers.location) {
+          return resolve(this._fetchJson(res.headers.location));
+        }
+        let data = '';
+        res.on('data', c => (data += c));
+        res.on('end', () => {
+          try { resolve(JSON.parse(data)); } catch { reject(new Error('Invalid JSON')); }
+        });
+      }).on('error', reject);
+    });
+  }
+
+  _fetchText(url) {
+    return new Promise((resolve, reject) => {
+      https.get(url, { headers: { 'User-Agent': 'booklib-init/1.0' } }, res => {
+        if (res.statusCode >= 300 && res.statusCode < 400 && res.headers.location) {
+          return resolve(this._fetchText(res.headers.location));
+        }
+        if (res.statusCode !== 200) {
+          reject(new Error(`HTTP ${res.statusCode} for ${url}`));
+          res.resume();
+          return;
+        }
+        let data = '';
+        res.on('data', c => (data += c));
+        res.on('end', () => resolve(data));
+      }).on('error', reject);
+    });
+  }
+}