Linux CLI Tool

Total Recall for your AI coding sessions

Search across every Claude Code and Codex CLI conversation you've ever had. Find that command, that fix, that solution from weeks ago in seconds.

Claude Code (Anthropic) Codex CLI (OpenAI)

Install Now


              $ /total-recall how did i fix the auth bug?


              ⏺ Skill(total-recall)

                ⎿ Successfully loaded skill


              ⏺ Contextify Total Recall


                Found: Auth bug fix from Jan 8, 2026


                Summary:

                - Race condition in token refresh

                - Fixed with mutex lock in AuthManager.swift:142

                - Committed as "fix(auth): prevent token refresh race"

An ingestion engine for your AI conversations.

The Linux CLI indexes your Claude Code and Codex transcripts into a single searchable database. Access them via the command-line tool, the Total Recall skill, or a researcher subagent in either Claude Code or Codex.

Note: Real-time summarization and the visual conversation timeline are macOS-only features. The Linux CLI focuses on indexing and search.

Built for Linux power users

🔍

Full-Text Search

Instant search across all your conversations. Find solutions by keyword, project, or time range without remembering exact phrasing.

📂

Cross-Project Queries

Search across every project at once, or filter to a specific repo. Your entire AI coding history in one place.

🌳

Git Worktree Support

First-class support for git worktrees. Contextify understands your multi-branch workflow and keeps context organized.

Complex searches run in the background

For multi-step research, delegate to the researcher subagent. In Claude Code it is query:contextify-researcher (shipped via the plugin). In Codex it is contextify_researcher (installed to $CODEX_HOME/agents/). Both run multiple queries in a read-only sandbox and return a cited synthesis.


              > Use a task to figure out what decisions we made about the database schema


              ● query:contextify-researcher(Search for database schema decisions)

                ⎿ Done (9 tool uses · 34.1k tokens · 1m 14s)


              ⏺ Contextify Total Recall


                Found: 4 conversations about schema design


                Key decisions:

                - Switched from JSON to normalized tables (Jan 3)

                - Added composite index on project_id + timestamp

                - Kept content as TEXT, not BLOB

Get Started

One-line install

curl -fsSL https://contextify.sh/install.sh | sh

Detects your architecture (x86_64 or ARM64), installs the binary, ingests existing transcripts, and sets up automatic ingestion via systemd.

The installer will:

Download the latest contextify binary for your architecture
Install to ~/.local/bin/ (XDG-compliant)
Install the Total Recall skill for Claude Code and Codex
Ingest any existing transcripts from ~/.claude/ and ~/.codex/
Set up a systemd user timer for automatic periodic ingestion

Or download manually

Intel / AMD

Most servers and desktops

Download x86_64

ARM64

Raspberry Pi, AWS Graviton

Download ARM64

Verify and use


              # Check installation

              $ contextify status


              # Run health check

              $ contextify doctor


              # Then in Claude Code or Codex, use the skill:

              $ /total-recall what did we build yesterday?

Help and Support

For setup guides, periodic ingestion, and troubleshooting:

Documentation Contact Us

Requirements

Linux x86_64 or ARM64

Ubuntu, Debian, Fedora, Arch, etc.

Claude Code or Codex CLI

At least one AI coding tool installed

The CLI tool reads conversation transcripts from ~/.claude/ and ~/.codex/