# LLM-Wiki for Personal Agent

A personal-agent memory system that turns daily conversations into a living context graph, helping agents build a more useful knowledge model over time.

Canonical: https://jonathantipper.com/projects/llm-wiki-hermes-agent
JSON: https://jonathantipper.com/projects/llm-wiki-hermes-agent.json
Status: building
Started: 2026
Tags: agents, memory, knowledge-graph, personal-infrastructure

## What it is

LLM-Wiki for Personal Agent is a dynamic context graph created by a personal agent running on Hermes.

The graph grows from daily conversations and turns recurring context into structured knowledge: projects, routines, values, interests, constraints, people, decisions, and working preferences.

## Why it matters

The goal is not just recall. It is usefulness.

As the agent learns more about work patterns, active projects, recurring context, and preferences, the graph becomes a knowledge layer for the broader agent system. The aim is to make future agents less generic and more context-aware without needing every conversation to start from zero.

## How it fits

This sits close to OpenClaw Agents, Launchpad, and the broader home infrastructure work.

OpenClaw is the fleet. Hermes is the personal-agent layer. LLM-Wiki is the memory and context graph that helps those agents understand the person and environment they are working with.

## Current status

Building. The current graph view shows the emerging shape of the personal context layer.