Cursor

Cursor stores agent transcripts as JSONL under ~/.cursor/projects, plus VS Code-family SQLite chats and a local full-text search index. IRFlow merges those into one AI Query History tab.

Back to AI Query History.

Canonical paths

Platform	Path
all	$CURSOR_AGENT_HOME or ~/.cursor/projects/<slug>/agent-transcripts/
macOS	~/Library/Application Support/Cursor/User/globalStorage/conversation-search.db
Windows	%APPDATA%\Cursor\User\globalStorage\conversation-search.db
Linux	~/.config/Cursor/User/globalStorage/conversation-search.db
all	globalStorage/state.vscdb, workspaceStorage/*/state.vscdb, ~/.cursor/chats/**/store.db

Typical transcript path:

~/.cursor/projects/<project-slug>/agent-transcripts/<session-id>/<session-id>.jsonl

Windows: %USERPROFILE%\.cursor\projects\...

What IRFlow extracts

Artifact	Status	Why it matters
~/.cursor/projects/<slug>/agent-transcripts/**/*.{jsonl,txt}	Parsed	Instructions, responses, tool-use text, sidechain flags, workspace attribution.
globalStorage/state.vscdb, workspaceStorage/*/state.vscdb, ~/.cursor/chats/**/store.db	Parsed when SQLite support is available	Composer/global/workspace chats that are not in the transcript files.
Cursor/User/globalStorage/conversation-search.db plus WAL/SHM	Parsed	Local FTS index: title, indexed body, conversation ID, source/scope, archive state, update time.

Each JSONL line is a user or assistant message with structured message.content blocks (text, tool calls). IRFlow uses embedded timestamp / createdAt values when present; file birth/mtime spreading is only a fallback for rows without per-message time. Project slugs under projects/ decode to filesystem paths when possible.

conversation-search.db is also accepted as a standalone artifact or through its parent Cursor User folder. Each indexed conversation becomes a searchable row with title in Summary, indexed body in FullText, conversation ID in SessionId, and the recorded update time.

SQLite sources are snapshotted with available WAL/SHM companions so recent composer/search rows are not silently missed.

How to import

1. File → Open… and select ~/.cursor, the Cursor User folder, or conversation-search.db, or Tools → Analysis → AI Artifacts → AI Apps → Cursor…
2. Agent transcripts and available Cursor SQLite stores consolidate into one tab.
3. Subagent folders (subagents/) and isSidechain transcript lines are skipped unless you choose Include subagents.

Collect AI Artifacts also finds Cursor roots on This Mac or inside a KAPE/triage folder (Windows/macOS/Linux profile layouts).

Investigation tips

• Prefer per-message createdAt / timestamp from JSONL and composer DB. File-mtime spread is a last resort.
• Sidechain / subagent lines are easy to over-collect on a large developer workstation — start with main sessions.
• conversation-search.db can still hold indexed body text after a transcript file is gone. Compare SessionId against recovered JSONL.

Limitations

• Composer/workspace recovery requires SQLite support (better-sqlite3 rebuilt for Electron).
• Proprietary or encrypted Cursor stores beyond the paths above are not claimed.

See also

• AI Query History overview
• Gemini
• AI Artifacts