@mrpatronz/nexusflow 0.2.17 → 0.2.19

package/src/utils/session-finder.ts

@@ -1,483 +1,483 @@
-/**
- * @module utils/session-finder
- * Scans local storage directories of various AI coding assistants to find conversation histories.
- */
-
-import * as fs from 'node:fs/promises';
-import * as path from 'node:path';
-import * as os from 'node:os';
-import type { AISession, ChatMessage } from '../types.js';
-
-/**
- * Derives the directory name that Claude Code uses inside `~/.claude/projects/`
- * for a given absolute project directory path.
- * It replaces special characters /, \, space, -, and _ with a hyphen.
- */
-export function getClaudeProjectFolderName(absolutePath: string): string {
-  return absolutePath.replace(/[\\\/ _-]+/g, '-');
-}
-
-/**
- * Recursively retrieves all files in a directory that match a specific extension.
- */
-async function getFilesRecursively(dir: string, extension: string): Promise<string[]> {
-  const result: string[] = [];
-  try {
-    const entries = await fs.readdir(dir, { withFileTypes: true });
-    for (const entry of entries) {
-      const fullPath = path.join(dir, entry.name);
-      if (entry.isDirectory()) {
-        const subFiles = await getFilesRecursively(fullPath, extension);
-        result.push(...subFiles);
-      } else if (entry.isFile() && entry.name.endsWith(extension)) {
-        result.push(fullPath);
-      }
-    }
-  } catch {}
-  return result;
-}
-
-/**
- * Scans the local filesystem for conversation histories belonging to Claude, Antigravity,
- * Codex, and Copilot that relate to the specified workspace.
- *
- * @param workspacePath - Root directory of the active workspace.
- * @param repoPaths - Directories of sub-repositories included in the workspace.
- * @returns A promise that resolves to an array of {@link AISession} objects sorted by update time descending.
- */
-export async function findSessions(workspacePath: string, repoPaths: string[] = []): Promise<AISession[]> {
-  const sessions: AISession[] = [];
-  const wsFolderName = path.basename(workspacePath);
-
-  // Normalization helper for accurate path comparison
-  const normalizePath = (p: string) => path.normalize(p).toLowerCase();
-  const normWorkspace = normalizePath(workspacePath);
-  const normRepos = repoPaths.map(r => normalizePath(r));
-
-  const isPathMatch = (sessPath?: string) => {
-    if (!sessPath) return false;
-    const normSess = normalizePath(sessPath);
-    if (normSess === normWorkspace || normSess.startsWith(normWorkspace + path.sep)) return true;
-    for (const r of normRepos) {
-      if (normSess === r || normSess.startsWith(r + path.sep)) return true;
-    }
-    return false;
-  };
-
-  // ─── 1. Scan Antigravity Sessions ──────────────────────────────────────
-  const agDir = path.join(os.homedir(), '.gemini', 'antigravity-cli');
-  const agHistoryPath = path.join(agDir, 'history.jsonl');
-  
-  try {
-    const historyContent = await fs.readFile(agHistoryPath, 'utf-8');
-    const lines = historyContent.split('\n').filter(Boolean);
-    const conversationsMap = new Map<string, { id: string; title: string; entries: any[] }>();
-
-    for (const line of lines) {
-      try {
-        const entry = JSON.parse(line);
-        if (entry.conversationId && isPathMatch(entry.workspace)) {
-          if (!conversationsMap.has(entry.conversationId)) {
-            conversationsMap.set(entry.conversationId, {
-              id: entry.conversationId,
-              title: entry.display || 'Untitled Conversation',
-              entries: [],
-            });
-          }
-          conversationsMap.get(entry.conversationId)!.entries.push(entry);
-        }
-      } catch {}
-    }
-
-    for (const [convId, convData] of conversationsMap.entries()) {
-      const transcriptPath = path.join(agDir, 'brain', convId, '.system_generated', 'logs', 'transcript.jsonl');
-      let messageCount = 0;
-      let createdAt = new Date(convData.entries[0]?.timestamp || Date.now()).toISOString();
-      let updatedAt = new Date(convData.entries[convData.entries.length - 1]?.timestamp || Date.now()).toISOString();
-      let title = convData.title;
-
-      try {
-        const transcriptContent = await fs.readFile(transcriptPath, 'utf-8');
-        const transcriptLines = transcriptContent.split('\n').filter(Boolean);
-        let firstUserMessage = null;
-
-        for (const line of transcriptLines) {
-          const tObj = JSON.parse(line);
-          if (tObj.type === 'USER_INPUT' || tObj.type === 'PLANNER_RESPONSE') {
-            messageCount++;
-            if (tObj.type === 'USER_INPUT' && !firstUserMessage) {
-              firstUserMessage = tObj.content;
-            }
-          }
-        }
-
-        if (firstUserMessage) {
-          const cleanMatch = firstUserMessage.match(/<USER_REQUEST>([\s\S]*?)<\/USER_REQUEST>/);
-          title = cleanMatch ? cleanMatch[1].trim() : firstUserMessage.trim();
-        }
-      } catch {
-        messageCount = convData.entries.length * 2;
-      }
-
-      if (title.length > 80) {
-        title = title.substring(0, 80) + '...';
-      }
-
-      sessions.push({
-        id: convId,
-        assistant: 'antigravity',
-        title,
-        createdAt,
-        updatedAt,
-        messageCount,
-        workspacePath,
-      });
-    }
-  } catch {}
-
-  // ─── 2. Scan Claude Code Sessions ──────────────────────────────────────
-  const claudeConfigDir = process.env.CLAUDE_CONFIG_DIR || path.join(os.homedir(), '.claude');
-  const claudeProjectsDir = path.join(claudeConfigDir, 'projects');
-  
-  try {
-    const candidateFolders = [
-      getClaudeProjectFolderName(workspacePath),
-      ...repoPaths.map(r => getClaudeProjectFolderName(r))
-    ];
-
-    for (const folder of candidateFolders) {
-      const projectPath = path.join(claudeProjectsDir, folder);
-      try {
-        const files = await fs.readdir(projectPath);
-        const jsonlFiles = files.filter(f => f.endsWith('.jsonl'));
-
-        for (const file of jsonlFiles) {
-          const filePath = path.join(projectPath, file);
-          const sessionId = path.basename(file, '.jsonl');
-
-          try {
-            const content = await fs.readFile(filePath, 'utf-8');
-            const lines = content.split('\n').filter(Boolean);
-            
-            let messageCount = 0;
-            let createdAt = null;
-            let updatedAt = null;
-            let title = 'Claude Session';
-
-            for (const line of lines) {
-              const record = JSON.parse(line);
-              if (record.timestamp) {
-                if (!createdAt) createdAt = record.timestamp;
-                updatedAt = record.timestamp;
-              }
-
-              if (record.type === 'user' || record.type === 'assistant') {
-                messageCount++;
-                if (record.type === 'user' && title === 'Claude Session') {
-                  const rawText = record.message?.text || record.message?.content || '';
-                  if (typeof rawText === 'string') {
-                    title = rawText.trim();
-                  } else if (Array.isArray(rawText)) {
-                    title = rawText.map((b: any) => b.text || b.content || '').join(' ').trim();
-                  }
-                }
-              }
-            }
-
-            if (title.length > 80) {
-              title = title.substring(0, 80) + '...';
-            }
-
-            sessions.push({
-              id: sessionId,
-              assistant: 'claude',
-              title,
-              createdAt: createdAt || new Date().toISOString(),
-              updatedAt: updatedAt || new Date().toISOString(),
-              messageCount,
-              workspacePath,
-            });
-          } catch {}
-        }
-      } catch {}
-    }
-  } catch {}
-
-  // ─── 3. Scan OpenAI Codex Sessions ──────────────────────────────────────
-  const codexDir = path.join(os.homedir(), '.codex');
-  const codexSessionsDir = path.join(codexDir, 'sessions');
-  try {
-    const codexFiles = await getFilesRecursively(codexSessionsDir, '.jsonl');
-    for (const file of codexFiles) {
-      try {
-        const content = await fs.readFile(file, 'utf-8');
-        // Heuristic: check if the file references our workspace path or folder name
-        if (!content.toLowerCase().includes(normWorkspace) && !content.includes(wsFolderName)) {
-          continue;
-        }
-
-        const lines = content.split('\n').filter(Boolean);
-        let messageCount = 0;
-        let createdAt: string | null = null;
-        let updatedAt: string | null = null;
-        let title = 'Codex Session';
-
-        for (const line of lines) {
-          try {
-            const record = JSON.parse(line);
-            if (record.timestamp) {
-              if (!createdAt) createdAt = new Date(record.timestamp).toISOString();
-              updatedAt = new Date(record.timestamp).toISOString();
-            }
-            const role = record.role || record.message?.role || (record.prompt ? 'user' : record.completion ? 'assistant' : null);
-            if (role === 'user' || role === 'assistant') {
-              messageCount++;
-              if (role === 'user' && title === 'Codex Session') {
-                const text = record.content || record.prompt || record.message?.content || '';
-                if (text) title = text.trim();
-              }
-            }
-          } catch {}
-        }
-
-        const sessionId = path.basename(file, '.jsonl');
-        sessions.push({
-          id: sessionId,
-          assistant: 'codex',
-          title: title.length > 80 ? title.substring(0, 80) + '...' : title,
-          createdAt: createdAt || new Date().toISOString(),
-          updatedAt: updatedAt || new Date().toISOString(),
-          messageCount,
-          workspacePath,
-        });
-      } catch {}
-    }
-  } catch {}
-
-  // ─── 4. Scan GitHub Copilot Sessions ────────────────────────────────────
-  const copilotDir = path.join(os.homedir(), '.copilot');
-  const copilotSessionsDirs = [
-    path.join(copilotDir, 'session-state'),
-    path.join(copilotDir, 'history-session-state')
-  ];
-  for (const dir of copilotSessionsDirs) {
-    try {
-      const copilotFiles = await getFilesRecursively(dir, '.json');
-      for (const file of copilotFiles) {
-        try {
-          const content = await fs.readFile(file, 'utf-8');
-          // Heuristic check
-          if (!content.toLowerCase().includes(normWorkspace) && !content.includes(wsFolderName)) {
-            continue;
-          }
-
-          const record = JSON.parse(content);
-          let messageCount = 0;
-          let title = 'Copilot Session';
-          let messagesList: any[] = [];
-
-          if (Array.isArray(record)) {
-            messagesList = record;
-          } else if (record && Array.isArray(record.messages)) {
-            messagesList = record.messages;
-          }
-
-          for (const msg of messagesList) {
-            const role = msg.role || (msg.type === 'user' ? 'user' : msg.type === 'assistant' ? 'assistant' : null);
-            if (role === 'user' || role === 'assistant') {
-              messageCount++;
-              if (role === 'user' && title === 'Copilot Session') {
-                const text = msg.content || msg.text || '';
-                if (text) title = text.trim();
-              }
-            }
-          }
-
-          const stats = await fs.stat(file);
-          const sessionId = path.basename(file, '.json');
-
-          sessions.push({
-            id: sessionId,
-            assistant: 'copilot',
-            title: title.length > 80 ? title.substring(0, 80) + '...' : title,
-            createdAt: stats.birthtime.toISOString(),
-            updatedAt: stats.mtime.toISOString(),
-            messageCount,
-            workspacePath,
-          });
-        } catch {}
-      }
-    } catch {}
-  }
-
-  // Sort by updatedAt descending
-  sessions.sort((a, b) => new Date(b.updatedAt).getTime() - new Date(a.updatedAt).getTime());
-  return sessions;
-}
-
-/**
- * Reads and parses the full transcript for a specific assistant conversation session.
- *
- * @param assistant - The name of the AI assistant ('antigravity', 'claude', 'codex', 'copilot').
- * @param sessionId - Unique ID of the conversation session.
- * @returns A promise that resolves to an array of {@link ChatMessage} objects.
- */
-export async function getSessionTranscript(assistant: string, sessionId: string): Promise<ChatMessage[]> {
-  const messages: ChatMessage[] = [];
-
-  if (assistant === 'antigravity') {
-    const agDir = path.join(os.homedir(), '.gemini', 'antigravity-cli');
-    const transcriptPath = path.join(agDir, 'brain', sessionId, '.system_generated', 'logs', 'transcript.jsonl');
-    const content = await fs.readFile(transcriptPath, 'utf-8');
-    const lines = content.split('\n').filter(Boolean);
-
-    for (const line of lines) {
-      const obj = JSON.parse(line);
-      if (obj.type === 'USER_INPUT') {
-        const rawContent = obj.content || '';
-        const cleanMatch = rawContent.match(/<USER_REQUEST>([\s\S]*?)<\/USER_REQUEST>/);
-        const cleanContent = cleanMatch ? cleanMatch[1].trim() : rawContent.trim();
-        messages.push({
-          role: 'user',
-          content: cleanContent,
-          timestamp: obj.created_at || obj.timestamp,
-        });
-      } else if (obj.type === 'PLANNER_RESPONSE') {
-        messages.push({
-          role: 'assistant',
-          content: (obj.content || '').trim(),
-          timestamp: obj.created_at || obj.timestamp,
-        });
-      }
-    }
-  } else if (assistant === 'claude') {
-    const claudeConfigDir = process.env.CLAUDE_CONFIG_DIR || path.join(os.homedir(), '.claude');
-    const claudeProjectsDir = path.join(claudeConfigDir, 'projects');
-    const projectFolders = await fs.readdir(claudeProjectsDir).catch(() => [] as string[]);
-    let transcriptPath: string | null = null;
-
-    for (const folder of projectFolders) {
-      const p = path.join(claudeProjectsDir, folder, `${sessionId}.jsonl`);
-      try {
-        await fs.access(p);
-        transcriptPath = p;
-        break;
-      } catch {}
-    }
-
-    if (!transcriptPath) {
-      throw new Error(`Claude session ${sessionId} not found`);
-    }
-
-    const content = await fs.readFile(transcriptPath, 'utf-8');
-    const lines = content.split('\n').filter(Boolean);
-
-    for (const line of lines) {
-      const record = JSON.parse(line);
-      if (record.type === 'user' || record.role === 'user') {
-        const rawText = record.message?.text || record.message?.content || record.text || '';
-        let cleanText = '';
-        if (typeof rawText === 'string') {
-          cleanText = rawText.trim();
-        } else if (Array.isArray(rawText)) {
-          cleanText = rawText.map((b: any) => b.text || b.content || '').join(' ').trim();
-        }
-        messages.push({
-          role: 'user',
-          content: cleanText,
-          timestamp: record.timestamp || record.created_at,
-        });
-      } else if (record.type === 'assistant' || record.role === 'assistant') {
-        const rawText = record.message?.text || record.message?.content || record.text || '';
-        let cleanText = '';
-        if (typeof rawText === 'string') {
-          cleanText = rawText.trim();
-        } else if (Array.isArray(rawText)) {
-          cleanText = rawText.map((b: any) => b.text || b.content || '').join(' ').trim();
-        }
-        messages.push({
-          role: 'assistant',
-          content: cleanText,
-          timestamp: record.timestamp || record.created_at,
-        });
-      }
-    }
-  } else if (assistant === 'codex') {
-    const codexDir = path.join(os.homedir(), '.codex');
-    const codexSessionsDir = path.join(codexDir, 'sessions');
-    const codexFiles = await getFilesRecursively(codexSessionsDir, '.jsonl');
-    let transcriptPath: string | null = null;
-
-    for (const file of codexFiles) {
-      if (path.basename(file, '.jsonl') === sessionId) {
-        transcriptPath = file;
-        break;
-      }
-    }
-
-    if (!transcriptPath) {
-      throw new Error(`Codex session ${sessionId} not found`);
-    }
-
-    const content = await fs.readFile(transcriptPath, 'utf-8');
-    const lines = content.split('\n').filter(Boolean);
-
-    for (const line of lines) {
-      const record = JSON.parse(line);
-      const role = record.role || record.message?.role || (record.prompt ? 'user' : record.completion ? 'assistant' : null);
-      if (role === 'user' || role === 'assistant') {
-        const text = record.content || record.prompt || record.completion || record.message?.content || '';
-        messages.push({
-          role: role as 'user' | 'assistant',
-          content: text.trim(),
-          timestamp: record.timestamp ? new Date(record.timestamp).toISOString() : undefined,
-        });
-      }
-    }
-  } else if (assistant === 'copilot') {
-    const copilotDir = path.join(os.homedir(), '.copilot');
-    const copilotSessionsDirs = [
-      path.join(copilotDir, 'session-state'),
-      path.join(copilotDir, 'history-session-state')
-    ];
-    let transcriptPath: string | null = null;
-
-    for (const dir of copilotSessionsDirs) {
-      const files = await getFilesRecursively(dir, '.json');
-      for (const file of files) {
-        if (path.basename(file, '.json') === sessionId) {
-          transcriptPath = file;
-          break;
-        }
-      }
-      if (transcriptPath) break;
-    }
-
-    if (!transcriptPath) {
-      throw new Error(`Copilot session ${sessionId} not found`);
-    }
-
-    const content = await fs.readFile(transcriptPath, 'utf-8');
-    const record = JSON.parse(content);
-    let messagesList: any[] = [];
-
-    if (Array.isArray(record)) {
-      messagesList = record;
-    } else if (record && Array.isArray(record.messages)) {
-      messagesList = record.messages;
-    }
-
-    for (const msg of messagesList) {
-      const role = msg.role || (msg.type === 'user' ? 'user' : msg.type === 'assistant' ? 'assistant' : null);
-      if (role === 'user' || role === 'assistant') {
-        messages.push({
-          role: role as 'user' | 'assistant',
-          content: (msg.content || msg.text || '').trim(),
-          timestamp: msg.timestamp || msg.created_at,
-        });
-      }
-    }
-  }
-
-  return messages;
-}
+/**
+ * @module utils/session-finder
+ * Scans local storage directories of various AI coding assistants to find conversation histories.
+ */
+
+import * as fs from 'node:fs/promises';
+import * as path from 'node:path';
+import * as os from 'node:os';
+import type { AISession, ChatMessage } from '../types.js';
+
+/**
+ * Derives the directory name that Claude Code uses inside `~/.claude/projects/`
+ * for a given absolute project directory path.
+ * It replaces special characters /, \, space, -, and _ with a hyphen.
+ */
+export function getClaudeProjectFolderName(absolutePath: string): string {
+  return absolutePath.replace(/[\\\/ _-]+/g, '-');
+}
+
+/**
+ * Recursively retrieves all files in a directory that match a specific extension.
+ */
+async function getFilesRecursively(dir: string, extension: string): Promise<string[]> {
+  const result: string[] = [];
+  try {
+    const entries = await fs.readdir(dir, { withFileTypes: true });
+    for (const entry of entries) {
+      const fullPath = path.join(dir, entry.name);
+      if (entry.isDirectory()) {
+        const subFiles = await getFilesRecursively(fullPath, extension);
+        result.push(...subFiles);
+      } else if (entry.isFile() && entry.name.endsWith(extension)) {
+        result.push(fullPath);
+      }
+    }
+  } catch {}
+  return result;
+}
+
+/**
+ * Scans the local filesystem for conversation histories belonging to Claude, Antigravity,
+ * Codex, and Copilot that relate to the specified workspace.
+ *
+ * @param workspacePath - Root directory of the active workspace.
+ * @param repoPaths - Directories of sub-repositories included in the workspace.
+ * @returns A promise that resolves to an array of {@link AISession} objects sorted by update time descending.
+ */
+export async function findSessions(workspacePath: string, repoPaths: string[] = []): Promise<AISession[]> {
+  const sessions: AISession[] = [];
+  const wsFolderName = path.basename(workspacePath);
+
+  // Normalization helper for accurate path comparison
+  const normalizePath = (p: string) => path.normalize(p).toLowerCase();
+  const normWorkspace = normalizePath(workspacePath);
+  const normRepos = repoPaths.map(r => normalizePath(r));
+
+  const isPathMatch = (sessPath?: string) => {
+    if (!sessPath) return false;
+    const normSess = normalizePath(sessPath);
+    if (normSess === normWorkspace || normSess.startsWith(normWorkspace + path.sep)) return true;
+    for (const r of normRepos) {
+      if (normSess === r || normSess.startsWith(r + path.sep)) return true;
+    }
+    return false;
+  };
+
+  // ─── 1. Scan Antigravity Sessions ──────────────────────────────────────
+  const agDir = path.join(os.homedir(), '.gemini', 'antigravity-cli');
+  const agHistoryPath = path.join(agDir, 'history.jsonl');
+  
+  try {
+    const historyContent = await fs.readFile(agHistoryPath, 'utf-8');
+    const lines = historyContent.split('\n').filter(Boolean);
+    const conversationsMap = new Map<string, { id: string; title: string; entries: any[] }>();
+
+    for (const line of lines) {
+      try {
+        const entry = JSON.parse(line);
+        if (entry.conversationId && isPathMatch(entry.workspace)) {
+          if (!conversationsMap.has(entry.conversationId)) {
+            conversationsMap.set(entry.conversationId, {
+              id: entry.conversationId,
+              title: entry.display || 'Untitled Conversation',
+              entries: [],
+            });
+          }
+          conversationsMap.get(entry.conversationId)!.entries.push(entry);
+        }
+      } catch {}
+    }
+
+    for (const [convId, convData] of conversationsMap.entries()) {
+      const transcriptPath = path.join(agDir, 'brain', convId, '.system_generated', 'logs', 'transcript.jsonl');
+      let messageCount = 0;
+      let createdAt = new Date(convData.entries[0]?.timestamp || Date.now()).toISOString();
+      let updatedAt = new Date(convData.entries[convData.entries.length - 1]?.timestamp || Date.now()).toISOString();
+      let title = convData.title;
+
+      try {
+        const transcriptContent = await fs.readFile(transcriptPath, 'utf-8');
+        const transcriptLines = transcriptContent.split('\n').filter(Boolean);
+        let firstUserMessage = null;
+
+        for (const line of transcriptLines) {
+          const tObj = JSON.parse(line);
+          if (tObj.type === 'USER_INPUT' || tObj.type === 'PLANNER_RESPONSE') {
+            messageCount++;
+            if (tObj.type === 'USER_INPUT' && !firstUserMessage) {
+              firstUserMessage = tObj.content;
+            }
+          }
+        }
+
+        if (firstUserMessage) {
+          const cleanMatch = firstUserMessage.match(/<USER_REQUEST>([\s\S]*?)<\/USER_REQUEST>/);
+          title = cleanMatch ? cleanMatch[1].trim() : firstUserMessage.trim();
+        }
+      } catch {
+        messageCount = convData.entries.length * 2;
+      }
+
+      if (title.length > 80) {
+        title = title.substring(0, 80) + '...';
+      }
+
+      sessions.push({
+        id: convId,
+        assistant: 'antigravity',
+        title,
+        createdAt,
+        updatedAt,
+        messageCount,
+        workspacePath,
+      });
+    }
+  } catch {}
+
+  // ─── 2. Scan Claude Code Sessions ──────────────────────────────────────
+  const claudeConfigDir = process.env.CLAUDE_CONFIG_DIR || path.join(os.homedir(), '.claude');
+  const claudeProjectsDir = path.join(claudeConfigDir, 'projects');
+  
+  try {
+    const candidateFolders = [
+      getClaudeProjectFolderName(workspacePath),
+      ...repoPaths.map(r => getClaudeProjectFolderName(r))
+    ];
+
+    for (const folder of candidateFolders) {
+      const projectPath = path.join(claudeProjectsDir, folder);
+      try {
+        const files = await fs.readdir(projectPath);
+        const jsonlFiles = files.filter(f => f.endsWith('.jsonl'));
+
+        for (const file of jsonlFiles) {
+          const filePath = path.join(projectPath, file);
+          const sessionId = path.basename(file, '.jsonl');
+
+          try {
+            const content = await fs.readFile(filePath, 'utf-8');
+            const lines = content.split('\n').filter(Boolean);
+            
+            let messageCount = 0;
+            let createdAt = null;
+            let updatedAt = null;
+            let title = 'Claude Session';
+
+            for (const line of lines) {
+              const record = JSON.parse(line);
+              if (record.timestamp) {
+                if (!createdAt) createdAt = record.timestamp;
+                updatedAt = record.timestamp;
+              }
+
+              if (record.type === 'user' || record.type === 'assistant') {
+                messageCount++;
+                if (record.type === 'user' && title === 'Claude Session') {
+                  const rawText = record.message?.text || record.message?.content || '';
+                  if (typeof rawText === 'string') {
+                    title = rawText.trim();
+                  } else if (Array.isArray(rawText)) {
+                    title = rawText.map((b: any) => b.text || b.content || '').join(' ').trim();
+                  }
+                }
+              }
+            }
+
+            if (title.length > 80) {
+              title = title.substring(0, 80) + '...';
+            }
+
+            sessions.push({
+              id: sessionId,
+              assistant: 'claude',
+              title,
+              createdAt: createdAt || new Date().toISOString(),
+              updatedAt: updatedAt || new Date().toISOString(),
+              messageCount,
+              workspacePath,
+            });
+          } catch {}
+        }
+      } catch {}
+    }
+  } catch {}
+
+  // ─── 3. Scan OpenAI Codex Sessions ──────────────────────────────────────
+  const codexDir = path.join(os.homedir(), '.codex');
+  const codexSessionsDir = path.join(codexDir, 'sessions');
+  try {
+    const codexFiles = await getFilesRecursively(codexSessionsDir, '.jsonl');
+    for (const file of codexFiles) {
+      try {
+        const content = await fs.readFile(file, 'utf-8');
+        // Heuristic: check if the file references our workspace path or folder name
+        if (!content.toLowerCase().includes(normWorkspace) && !content.includes(wsFolderName)) {
+          continue;
+        }
+
+        const lines = content.split('\n').filter(Boolean);
+        let messageCount = 0;
+        let createdAt: string | null = null;
+        let updatedAt: string | null = null;
+        let title = 'Codex Session';
+
+        for (const line of lines) {
+          try {
+            const record = JSON.parse(line);
+            if (record.timestamp) {
+              if (!createdAt) createdAt = new Date(record.timestamp).toISOString();
+              updatedAt = new Date(record.timestamp).toISOString();
+            }
+            const role = record.role || record.message?.role || (record.prompt ? 'user' : record.completion ? 'assistant' : null);
+            if (role === 'user' || role === 'assistant') {
+              messageCount++;
+              if (role === 'user' && title === 'Codex Session') {
+                const text = record.content || record.prompt || record.message?.content || '';
+                if (text) title = text.trim();
+              }
+            }
+          } catch {}
+        }
+
+        const sessionId = path.basename(file, '.jsonl');
+        sessions.push({
+          id: sessionId,
+          assistant: 'codex',
+          title: title.length > 80 ? title.substring(0, 80) + '...' : title,
+          createdAt: createdAt || new Date().toISOString(),
+          updatedAt: updatedAt || new Date().toISOString(),
+          messageCount,
+          workspacePath,
+        });
+      } catch {}
+    }
+  } catch {}
+
+  // ─── 4. Scan GitHub Copilot Sessions ────────────────────────────────────
+  const copilotDir = path.join(os.homedir(), '.copilot');
+  const copilotSessionsDirs = [
+    path.join(copilotDir, 'session-state'),
+    path.join(copilotDir, 'history-session-state')
+  ];
+  for (const dir of copilotSessionsDirs) {
+    try {
+      const copilotFiles = await getFilesRecursively(dir, '.json');
+      for (const file of copilotFiles) {
+        try {
+          const content = await fs.readFile(file, 'utf-8');
+          // Heuristic check
+          if (!content.toLowerCase().includes(normWorkspace) && !content.includes(wsFolderName)) {
+            continue;
+          }
+
+          const record = JSON.parse(content);
+          let messageCount = 0;
+          let title = 'Copilot Session';
+          let messagesList: any[] = [];
+
+          if (Array.isArray(record)) {
+            messagesList = record;
+          } else if (record && Array.isArray(record.messages)) {
+            messagesList = record.messages;
+          }
+
+          for (const msg of messagesList) {
+            const role = msg.role || (msg.type === 'user' ? 'user' : msg.type === 'assistant' ? 'assistant' : null);
+            if (role === 'user' || role === 'assistant') {
+              messageCount++;
+              if (role === 'user' && title === 'Copilot Session') {
+                const text = msg.content || msg.text || '';
+                if (text) title = text.trim();
+              }
+            }
+          }
+
+          const stats = await fs.stat(file);
+          const sessionId = path.basename(file, '.json');
+
+          sessions.push({
+            id: sessionId,
+            assistant: 'copilot',
+            title: title.length > 80 ? title.substring(0, 80) + '...' : title,
+            createdAt: stats.birthtime.toISOString(),
+            updatedAt: stats.mtime.toISOString(),
+            messageCount,
+            workspacePath,
+          });
+        } catch {}
+      }
+    } catch {}
+  }
+
+  // Sort by updatedAt descending
+  sessions.sort((a, b) => new Date(b.updatedAt).getTime() - new Date(a.updatedAt).getTime());
+  return sessions;
+}
+
+/**
+ * Reads and parses the full transcript for a specific assistant conversation session.
+ *
+ * @param assistant - The name of the AI assistant ('antigravity', 'claude', 'codex', 'copilot').
+ * @param sessionId - Unique ID of the conversation session.
+ * @returns A promise that resolves to an array of {@link ChatMessage} objects.
+ */
+export async function getSessionTranscript(assistant: string, sessionId: string): Promise<ChatMessage[]> {
+  const messages: ChatMessage[] = [];
+
+  if (assistant === 'antigravity') {
+    const agDir = path.join(os.homedir(), '.gemini', 'antigravity-cli');
+    const transcriptPath = path.join(agDir, 'brain', sessionId, '.system_generated', 'logs', 'transcript.jsonl');
+    const content = await fs.readFile(transcriptPath, 'utf-8');
+    const lines = content.split('\n').filter(Boolean);
+
+    for (const line of lines) {
+      const obj = JSON.parse(line);
+      if (obj.type === 'USER_INPUT') {
+        const rawContent = obj.content || '';
+        const cleanMatch = rawContent.match(/<USER_REQUEST>([\s\S]*?)<\/USER_REQUEST>/);
+        const cleanContent = cleanMatch ? cleanMatch[1].trim() : rawContent.trim();
+        messages.push({
+          role: 'user',
+          content: cleanContent,
+          timestamp: obj.created_at || obj.timestamp,
+        });
+      } else if (obj.type === 'PLANNER_RESPONSE') {
+        messages.push({
+          role: 'assistant',
+          content: (obj.content || '').trim(),
+          timestamp: obj.created_at || obj.timestamp,
+        });
+      }
+    }
+  } else if (assistant === 'claude') {
+    const claudeConfigDir = process.env.CLAUDE_CONFIG_DIR || path.join(os.homedir(), '.claude');
+    const claudeProjectsDir = path.join(claudeConfigDir, 'projects');
+    const projectFolders = await fs.readdir(claudeProjectsDir).catch(() => [] as string[]);
+    let transcriptPath: string | null = null;
+
+    for (const folder of projectFolders) {
+      const p = path.join(claudeProjectsDir, folder, `${sessionId}.jsonl`);
+      try {
+        await fs.access(p);
+        transcriptPath = p;
+        break;
+      } catch {}
+    }
+
+    if (!transcriptPath) {
+      throw new Error(`Claude session ${sessionId} not found`);
+    }
+
+    const content = await fs.readFile(transcriptPath, 'utf-8');
+    const lines = content.split('\n').filter(Boolean);
+
+    for (const line of lines) {
+      const record = JSON.parse(line);
+      if (record.type === 'user' || record.role === 'user') {
+        const rawText = record.message?.text || record.message?.content || record.text || '';
+        let cleanText = '';
+        if (typeof rawText === 'string') {
+          cleanText = rawText.trim();
+        } else if (Array.isArray(rawText)) {
+          cleanText = rawText.map((b: any) => b.text || b.content || '').join(' ').trim();
+        }
+        messages.push({
+          role: 'user',
+          content: cleanText,
+          timestamp: record.timestamp || record.created_at,
+        });
+      } else if (record.type === 'assistant' || record.role === 'assistant') {
+        const rawText = record.message?.text || record.message?.content || record.text || '';
+        let cleanText = '';
+        if (typeof rawText === 'string') {
+          cleanText = rawText.trim();
+        } else if (Array.isArray(rawText)) {
+          cleanText = rawText.map((b: any) => b.text || b.content || '').join(' ').trim();
+        }
+        messages.push({
+          role: 'assistant',
+          content: cleanText,
+          timestamp: record.timestamp || record.created_at,
+        });
+      }
+    }
+  } else if (assistant === 'codex') {
+    const codexDir = path.join(os.homedir(), '.codex');
+    const codexSessionsDir = path.join(codexDir, 'sessions');
+    const codexFiles = await getFilesRecursively(codexSessionsDir, '.jsonl');
+    let transcriptPath: string | null = null;
+
+    for (const file of codexFiles) {
+      if (path.basename(file, '.jsonl') === sessionId) {
+        transcriptPath = file;
+        break;
+      }
+    }
+
+    if (!transcriptPath) {
+      throw new Error(`Codex session ${sessionId} not found`);
+    }
+
+    const content = await fs.readFile(transcriptPath, 'utf-8');
+    const lines = content.split('\n').filter(Boolean);
+
+    for (const line of lines) {
+      const record = JSON.parse(line);
+      const role = record.role || record.message?.role || (record.prompt ? 'user' : record.completion ? 'assistant' : null);
+      if (role === 'user' || role === 'assistant') {
+        const text = record.content || record.prompt || record.completion || record.message?.content || '';
+        messages.push({
+          role: role as 'user' | 'assistant',
+          content: text.trim(),
+          timestamp: record.timestamp ? new Date(record.timestamp).toISOString() : undefined,
+        });
+      }
+    }
+  } else if (assistant === 'copilot') {
+    const copilotDir = path.join(os.homedir(), '.copilot');
+    const copilotSessionsDirs = [
+      path.join(copilotDir, 'session-state'),
+      path.join(copilotDir, 'history-session-state')
+    ];
+    let transcriptPath: string | null = null;
+
+    for (const dir of copilotSessionsDirs) {
+      const files = await getFilesRecursively(dir, '.json');
+      for (const file of files) {
+        if (path.basename(file, '.json') === sessionId) {
+          transcriptPath = file;
+          break;
+        }
+      }
+      if (transcriptPath) break;
+    }
+
+    if (!transcriptPath) {
+      throw new Error(`Copilot session ${sessionId} not found`);
+    }
+
+    const content = await fs.readFile(transcriptPath, 'utf-8');
+    const record = JSON.parse(content);
+    let messagesList: any[] = [];
+
+    if (Array.isArray(record)) {
+      messagesList = record;
+    } else if (record && Array.isArray(record.messages)) {
+      messagesList = record.messages;
+    }
+
+    for (const msg of messagesList) {
+      const role = msg.role || (msg.type === 'user' ? 'user' : msg.type === 'assistant' ? 'assistant' : null);
+      if (role === 'user' || role === 'assistant') {
+        messages.push({
+          role: role as 'user' | 'assistant',
+          content: (msg.content || msg.text || '').trim(),
+          timestamp: msg.timestamp || msg.created_at,
+        });
+      }
+    }
+  }
+
+  return messages;
+}