LLM-Wiki-Setup
Interview-driven setup kit: an AI agent builds you a provenance-tagged knowledge wiki + layered instruction stack for Claude Code, Codex, and free-claude-code, across any number of machines
Overview
Give your coding agent a memory and a map.
This is a setup kit that an AI agent (Claude Code, Codex, or free-claude-code) runs against your machine to build you a personal agentic work system:
It is built by interview, not template-filling: the agent first scans your machine (read-only, disclosed up front), infers what it can, and asks only what it can't. And it is not domain-locked: the wiki's page-type schema is composed per user from a catalog of types with a mechanism for minting new ones — it fits deep-learning research, product engineering, data work, ops, and long-form writing equally, including all of them at once.
The agent discloses what it will scan, interviews you in short adaptive waves, proposes a plan, and only then creates anything. Expect 15–40 minutes depending on how much of your existing world you want ingested on day one.
From the project’s README on GitHub.
At a glance
| Repository | Kentucky-Open-Science/LLM-Wiki-Setup |
|---|---|
| Research area | Other projects |
| Primary language | Python |
| Languages | Python 94%, Shell 6% |
| License | Apache-2.0 |
| Stars / forks | 1 / 0 |
| Open issues and pull requests | 0 |
| Created | 2026-08-24 |
| Last push | 2026-08-26 |
| Default branch | main |
What the README covers
- The problems this solves
- Quickstart
- What you end up with
- Supported agents
- What it will and won't do
- Agent directive
Top contributors
- @EvanDamron (3 commits)
Get the code
git clone https://github.com/Kentucky-Open-Science/LLM-Wiki-Setup.git