Contextify for Linux: Back Up Your AI Conversations and Make Them Searchable

Your AI conversations are stored in brittle .jsonl files

Every conversation you have with Claude Code or Codex CLI gets written to a .jsonl file on disk. These files are append-only logs with no indexing, no search, and no structure beyond raw JSON lines. They accumulate silently in ~/.claude/ and ~/.codex/, and if they get corrupted or deleted, those conversations are gone.

Contextify changes that. It reads those transcript files, parses them, and stores every message in a proper SQLite database with full-text search indexes. On macOS, this happens automatically in the background with a visual timeline. Now it works on Linux too.

One command to install

The installer detects your architecture (x86_64 or ARM64), downloads the binary, ingests any existing transcripts it finds, and sets up a systemd user timer for automatic periodic ingestion. If systemd is not available, it falls back to cron.

Once running, every new conversation you have with Claude Code or Codex gets indexed automatically.

The database powers Total Recall

The database Contextify builds is the same one that powers the /total-recall skill. Total Recall teaches your AI how to search your conversation history. Instead of trying to remember which session had that one useful solution, you just ask:

"Use total recall to find where we decided to use Redis instead of Postgres for the session cache."

The skill translates your question into targeted database queries and returns the relevant conversation excerpts. For more complex research, it spawns a contextify-researcher agent that runs multiple queries and synthesizes results across sessions.

This release also improves the skill itself: query expansion helps your AI construct better searches, JSON schema documentation reduces parsing errors, and better error handling means more reliable results.

Shared database, searchable from anywhere

If you store the database on Dropbox or iCloud Drive, you can use Contextify on your Mac to browse conversations visually and search on your Linux machine via the CLI. Same database, two interfaces.

Also in this release

Always-on-Top Window (macOS) - Contextify can now float above other windows. Toggle via Window > Float on Top. Useful for keeping the timeline visible while your AI works on a long task.

Unified CLI - The primary command is now just contextify (replacing contextify-query). Old name still works via symlink.

Questions or feedback? Email support or open an issue on GitHub.