LLM Wiki Newsroom workflow
A Claude Code template that builds a knowledge base from articles and PDFs: five newsroom roles and contradiction checks.
- 173 ★ on the repository
- Last commit
Extended access
Why this level
- The agent reads uploaded documents and rewrites wiki pages and Python scripts run locally.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add workflows/llm-wiki-newsroomDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The agent reads uploaded documents and rewrites wiki pages and Python scripts run locally.
This is third-party code. Review the repository files before installing.
What it does
LLM Wiki Newsroom is an open local knowledge base run by a multi-agent "newsroom" of five roles. Documents, articles and PDFs go into the `raw/` folder and `/wiki-ingest` makes the agent read them, extract entities, concepts and relationships and organize them into a cross-referenced wiki. It is a persistent alternative to RAG built on Karpathy's three-layer pattern: sources, wiki and agent operating rules. Roles divide the labor: the reporter gathers material and drafts, the columnist writes deep analysis, the copy editor checks formal rules in code with no model, the desk rereads drafts with fresh eyes and the editor-in-chief routes work. A page's author is never its reviewer and the rules self-update from mistakes. There are `/wiki-query`, `/wiki-lint`, `/wiki-graph`, `/wiki-news`, `/wiki-discover`, `/wiki-trail`, `/wiki-timeline` and an interactive knowledge graph.
Who it is for. Researchers, analysts and writers who accumulate documents on a topic and want a coherent wiki maintained by an agent.
Good fit when
- A folder of articles and PDFs on a topic has piled up and must become a linked wiki with cross-references.
- Contradictions between sources must be found and conflicting claims flagged.
- The wiki must answer a question grounded in pages with evidence links.
- You want to see unexpected connections through `/wiki-discover` or build a trail on a topic.
Not a fit when
- You need simple search across existing notes: the template builds a complex system of five roles.
- The agent is not Claude Code: Codex and Gemini CLI get only the basic workflow.
- There is no Python for the graph build and checks and access cannot be obtained.
Example request
Read all new documents from `raw/` with `/wiki-ingest raw/`, update related pages and show the contradictions found.Limitations
The full feature set, including cascading updates, a backlink index, contradiction tracking and associative discovery, works in Claude Code. Codex and Gemini CLI through `AGENTS.md` and `GEMINI.md` get only the basic workflow. Python with the packages from `requirements.txt` is needed, and if `igraph` and `leidenalg` have no wheels for your platform the two lines are deleted and the graph build falls back to pure Python. The graph is opened through a local web server and a double-click on `file://` is unsupported. The example corpus in the repository covers the debate over what open source means for AI.
How to disable. Delete the cloned `llm-wiki-newsroom` folder. The template leaves no background processes and the graph opens through a local server you start yourself.
FAQ
How do I install LLM Wiki Newsroom?
Run `git clone https://github.com/alfadur7/llm-wiki-newsroom.git`, enter the folder and run `pip install -r requirements.txt`. No keys or accounts are needed. You can also create your own repository with the "Use this template" button.
How do I add documents?
Put a file into `raw/` or a PDF into `raw/PDF/` and call `/wiki-ingest <file, folder or inbox>`. The agent reads them and refreshes about 10 to 15 related pages, skipping duplicates automatically.
Which commands does the template have?
Core: `/wiki-ingest`, `/wiki-query`, `/wiki-lint`, `/wiki-graph`. Extra: `/wiki-news`, `/wiki-export`, `/wiki-discover`, `/wiki-trail`, `/wiki-timeline`. Nine commands for Claude Code in total.
Does the template work with Codex and Gemini CLI?
It works in a basic mode: they lack slash commands and advanced features and the process runs on plain phrases through `AGENTS.md` and `GEMINI.md`. The full set is available in Claude Code.
How do I open the knowledge graph?
Start a local server with something like `python -m http.server` in the `graph/` folder and open `graph.html`. A double-click will not work because the page fetches its data.
Discussion
No questions or reviews yet.
Related
WorkflowsAn Obsidian vault template where an agent follows a CLAUDE.md schema to compile sources into a linked wiki and extend it.
LLM Wiki Manager keeps a personal Markdown wiki through an agent: it ingests sources, links pages and answers with citations
A workflow for Claude Code, Codex and Gemini CLI where the agent files source documents into a persistent, linked wiki
A CLI that creates an Obsidian-ready Markdown vault and installs four skills so an agent maintains the knowledge base.
A markdown research wiki template: the agent ingests sources, updates pages and checks every quote word for word against the file.
LLM Wiki
llm-wiki
Knowledge bases that AI compiles: topics, import, llm-wiki lint checks, the /wiki:ingest and /wiki:archive commands. A plugin for Claude Code and Codex
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| LLM Wiki Newsroom (this card) | no | 173 | |
| Karpathy LLM Wiki | no | 2,728 | a single-agent wiki on Karpathy's pattern; this one has five roles, a separate reviewer, a graph and associative discovery |
| LLM Wiki Compiler | no | 2,254 | a wiki compiler from sources; this one has a newsroom process with contradiction checks |
README in short
The template separates layers: `raw/` for sources, `wiki/` written only by the agent and rules in `CLAUDE.md` and `.claude/`. The example corpus is published as a GitHub Wiki.