al3rez

al3rez / zehn

Public

Fuzzy-find any prompt across claude, codex, pi & opencode histories, then resume that session. Fast Zig TUI.

10
1
89% credibility
Found May 29, 2026 at 13 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Zig
AI Summary

Zehn is a searchable history browser for AI coding assistants that reads prompts from Claude Code, Codex, pi, and opencode, displays them in a fuzzy-searchable list, and can resume any selected session with one keystroke.

How It Works

1
💭 You use multiple AI coding assistants

Over time, you switch between different AI helpers like Claude Code, Codex, pi, and opencode for different projects.

2
😕 You forget which one you asked

A week later, you want that thing you asked for, but you cannot remember which assistant you used or which project folder it was in.

3
🔍 You discover zehn

A small program that reads all your AI assistant histories at once and shows them in one searchable list.

4
⚡ You install it in one line

A single command downloads and sets up zehn on your computer, ready to search through all your past conversations.

5
⌨️ You search for what you need

You type a few letters and zehn instantly finds matching prompts from all your assistants, highlighted in their brand colors.

6
👆 You pick your prompt and press Enter

Using arrow keys to browse, you select the exact conversation you were looking for.

🎉 You're back in your session

zehn opens the right project folder and resumes your conversation exactly where you left off, as if no time passed.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 13 to 10 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is zehn?

Zig-based terminal tool that searches your AI coding agent history across Claude, Codex, pi, and opencode in one place. Type a few letters, find an old prompt, and hit Enter to drop back into that exact session in its original agent. Uses a Smith-Waterman-style fuzzy matching algorithm with word-boundary bonuses, so searching "auth" surfaces "add auth middleware" above scattered matches. Ships as a single binary with no runtime dependencies and parses about 1,300 sessions in 0.2 seconds on typical hardware.

Why is it gaining traction?

Most developers now bounce between multiple AI coding assistants, and weeks later you cannot remember which one you asked something to. Zehn solves this by pulling all histories into one searchable interface with agent-color-coded results. The matching is genuinely smart - it rewards word boundaries and consecutive matches, not just substring containment. It handles the SQLite database used by opencode gracefully, skipping it with a warning if sqlite3 is absent rather than failing. The interactive TUI feels snappy and professional.

Who should use this?

Developers who use Claude Code, Codex, pi, or opencode and have accumulated weeks or months of session history. If you find yourself re-explaining context to an agent because you cannot find what you asked before, this is for you. Power users who switch between agents will benefit most - once you see your entire prompt history in one view, you will not want to go back.

Verdict

Zehn solves a real problem for developers juggling multiple AI agents, with a fast, well-designed TUI that respects your time. With a credibility score of 0.8999999761581421%, the project is legitimate but early - only 10 stars and the codebase is still maturing. The fuzzy matching is rigorously tested against brute-force references, which inspires confidence despite the low star count. If you regularly switch between coding agents, install it and see if your prompt history reveals patterns you forgot. The risk is minimal and the potential time savings are significant.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.