recall is an advanced tool designed for efficiently searching your AI chat history. It supports interactions with various platforms, including Cursor, Claude Code, OpenAI Codex CLI, and pi. With a focus on speed and data integrity, recall creates a compact, searchable index of past conversations without moving, copying, or modifying your existing data.

Key Features

• Fast and Local: Quickly retrieve chat sessions directly from your native storage.
• Read-Only Access: Ensure that original data remains unchanged while allowing for instantaneous searches.
• Minimal Footprint: Utilizes a lightweight SQLite database that preserves details of your conversations without excess bloat.

How it Works

recall operates by indexing conversations stored natively in your tools. It builds a searchable index that includes excerpts and metadata, enabling users to query their chat history through the terminal. Here’s a glimpse of the interaction:

$ recall doctor
recall 0.1.0

sources:
  ✓ cursor  ~/Library/Application Support/Cursor/User/globalStorage/state.vscdb
  ✓ claude  ~/.claude/projects
  ✓ codex   ~/.codex/sessions

index: ~/.recall/index.sqlite (69.5 MiB)
  cursor  1514 sessions
  claude    25 sessions
  codex   1058 sessions
  total   2597 sessions

To search for specific topics, a simple query suffices:

$ recall "import cycle" --limit 3
2025-09-02 18:36  cursor  ~/code/acme-api    Fix import cycle in proto files
    id=cursor:94dc8775-5fd3-41e9-93d7-43d7dff795b6  msg=1  role=assistant

Integration with AI Agents

recall can seamlessly integrate into AI agent workflows. By adopting the Model Context Protocol (MCP), it enables agents to access past chat histories as needed, enhancing their capabilities:

For example, in Claude Code:

claude mcp add recall -- recall mcp

For Codex:

[mcp_servers.recall]
command = "recall"
args = ["mcp"]

Commands Overview

• recall index - Rebuild the index from all sources.
• recall <query> - Search for specific terms across all indexed sessions.
• recall last - Display the most recent session transcript.
• recall sessions - List recent chat sessions for easy access.

Data Sources Supported

recall reads from various storage solutions allowing seamless interaction with:

• Cursor
• Claude Code
• Codex CLI
• pi

The design philosophy behind recall emphasizes a read-only model ensuring the source remains trustworthy while providing a fast querying mechanism tailored for developers and researchers dealing with AI conversations. With its current version at v0.1, the roadmap includes incremental indexing and user-friendly interface enhancements.