klyne-ai

klyne-ai / klyne

Public

Local-first MCP server for Claude Code and Codex. Recovers post-/compact context, generates deterministic handoff prompts, and surfaces what's bloating your session. Reads the JSONL your AI already writes. No cloud.

13
0
100% credibility
Found May 11, 2026 at 13 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Go
AI Summary

Klyne is a local dashboard and toolkit that rescues lost context, tracks token usage, and provides proactive advice for AI coding sessions from Claude Code and Codex.

How It Works

1
🔍 Discover Klyne

You hear about Klyne, a helpful companion that saves and revives your AI coding chats when they get messy or hit limits.

2
📥 Get it running

Download and install with a simple command, like grabbing a favorite app.

3
🔗 Link your AI tools

Run one setup command to connect it to your coding AI sessions automatically.

4
🚀 Open your dashboard

Start the app and see a friendly web page listing all your recent AI chats.

5
💡 Smart advice appears

Get instant tips on saving context, spotting bloat, or handing off to a fresh chat without losing a beat.

6
📋 Recover or hand off

Click to rebuild lost details or copy a perfect summary to paste into your next AI session.

✅ Your sessions thrive

No more lost work or wasted limits—your AI coding flows smoothly forever.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 13 to 13 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 klyne?

klyne is a local-first Go tool that rescues Claude Code and Codex sessions by reading their JSONL logs directly—no cloud or proxies needed. It recovers context lost after /compact commands, generates deterministic handoff prompts for fresh chats, and surfaces bloating sessions with token timelines and health checks. Run it as a daemon for a web cockpit at localhost:7878, or use CLI commands like `klyne tokens` and MCP slash prompts like `/klyne:health`.

Why is it gaining traction?

Heavy Claude and Codex users hit rate limits and context bloat weekly; klyne fixes this with proactive advisor hooks that warn inline, reproducible proofs via `make proof`, and tools like `get_pre_compact_context` that pull lost details already in your JSONL. Unlike vague AI summaries, handoffs are structured Markdown—same input always yields identical output. Local-first on GitHub means zero telemetry and instant setup via brew.

Who should use this?

Claude Code power users debugging long sessions, Codex CLI devs tracking token spend across projects, or anyone hitting 5-hour windows and needing quick handoffs. Ideal for solo devs or small teams managing context drift without re-explaining file paths and commands every restart.

Verdict

Grab it if you're deep in Claude/Codex workflows—docs and proofs are solid, CLI/MCP integration seamless. At 13 stars and 1.0% credibility, it's early (low maturity, sparse adoption), but deterministic tools and audit commands build trust fast. Worth a `brew install` test run.

(187 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.