Texarkanine

OptMem Split — AI skill for Claude Code

AI community

Permanent memory for AI agents.

How to install OptMem Split

This entry records only its repository, not the path inside it, so there is no exact command to give. Open Texarkanine/OptMem-Split and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What OptMem Split does

Permanent memory for AI agents. A 400ish-token prompt, a script, plug and play.

Alternatives in AI

Pebrel — AI-native, GPU-accelerated terminal emulator for Windows with SSH, persistent sessions, split panes, and first 670 ★
Open Researcher — 🔥 Visual AI research assistant that displays real-time thinking, provides split-view analysis, and automatic 634 ★
Nebula — AI-native, GPU-accelerated terminal emulator for Windows with SSH, persistent sessions, split panes, and first 587 ★

README

OptMem

Permanent memory for AI agents. A 400-ish-token prompt, a script, plug and play.

This fork of [VictorTaelin/OptMem](https://github.com/VictorTaelin/OptMem) flips the default: **one memory store per project**, plus one global memory that follows you everywhere. Project first; `--global` for the rest.

![how OptMem works](anim/optmem.gif)

Install

curl -fsSL https://raw.githubusercontent.com/Texarkanine/OptMem-Split/main.split/install.sh | sh

It prints a `## Memory` block. Paste that at the top of your agent's `AGENTS.md` (or `CLAUDE.md`), and you are done. Run the same line again to update.

The tool lands at `~/.optmem/memo`; put `~/.optmem` on `PATH` to type `memo`.

Commands


`memo wake`	read both memories — global, then project; first command of every session
`memo note "..."`	record one memory: one line, up to 280 bytes (project by default)
`memo nap`	answer the merges that came due
`memo recall`	search every memory ever recorded, word for word
`memo zoom -`	open a tree node into its two halves
`memo forget -`	drop a bad summary; the next nap rebuilds it

Put `--global` before any command to reach the memory that follows you into every project. Merges arrive one at a time, in the output of `note`. Nothing ever runs in the background.

Why Split Memory

A single log is one identity, and wake spends its reading budget on the present. That is right for one continuous workstream and wrong for several: N interleaved projects make each one's detail decay at the rate of the other N−1, so a repo left alone for six months wakes up remembering nothing — its memories intact in the log and out of reach.

So every command speaks to the memory of the project in `$PWD`, keyed by the origin remote reduced to `owner/repo` (every worktree and host alias for one repo is one memory). `--global` reaches the one that follows you everywhere. `wake` alone reads both: who yo