PyGuy2000

K Mem — Development skill for Claude Code

Development community

Memory and governance for Claude Code sessions: a read-before-write gate, a context resolver with receipts, tiered project notes, handoffs between sessions, and docs checks that fail when the record a.

How to install K Mem

This entry records only its repository, not the path inside it, so there is no exact command to give. Open PyGuy2000/k-mem and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What K Mem does

Alternatives in Development

Resolver — Waza Skill Resolver 3.8k ★
Verify Theme — Run the full Graphite quality gate (drift, tokens, contrast, YAML) and report pass/fail 452 ★
Generate Release Notes — Generate user-focused release notes between two versions 112 ★

README

KazzerLabs K-mem: memory and governance for Claude Code sessions

[![CI](https://img.shields.io/github/actions/workflow/status/PyGuy2000/k-mem/ci.yml?branch=main&style=for-the-badge&label=CI&labelColor=080C16&color=00E1FF)](https://github.com/PyGuy2000/k-mem/actions/workflows/ci.yml) [![Release](https://img.shields.io/github/v/tag/PyGuy2000/k-mem?style=for-the-badge&label=release&labelColor=080C16&color=00E1FF)](https://github.com/PyGuy2000/k-mem/releases) [![License](https://img.shields.io/badge/license-MIT-FF941A?style=for-the-badge&labelColor=080C16)](LICENSE) [![Python](https://img.shields.io/badge/python-3.9%2B-00E1FF?style=for-the-badge&labelColor=080C16&logo=python&logoColor=white)](pyproject.toml)

[![Claude Code](https://img.shields.io/badge/Claude%20Code-plugin-FF941A?style=for-the-badge&labelColor=080C16)](https://code.claude.com/docs/en/plugins) [![Tests](https://img.shields.io/badge/tests-165-00E1FF?style=for-the-badge&labelColor=080C16)](tests) [![Dependencies](https://img.shields.io/badge/runtime%20deps-none-00E1FF?style=for-the-badge&labelColor=080C16)](plugins/k-mem/lib/kmem)

**[The problem](#the-problem)** · **[Install](#install)** · **[What you get](#what-you-get)** · **[First session](#the-first-session)** · **[Your own repo](#your-own-repo)** · **[Design notes](docs/design.md)**

The problem

Two things are true at once. Each session is a fresh model with no memory. The files are the memory. And a session can skip a file, or half-read it, and give a confident answer anyway. Nothing in the loop can tell the difference between "read it" and "said it read it". So the honest position is this: you cannot rely on me to be the check on me. The fix has to sit outside the agent, where I cannot skip it and cannot soften it.

That is Claude, on 2026-09-08, asked to design the thing that would constrain it. The wording is the assistant's own and it is treated here