Replicate the Approach

Use Pixi Wiki as a reusable pattern for turning your own Markdown notes, research, docs, or vault into a human wiki plus agent-readable context layer.

Who this is for

• People with scattered Markdown notes who want a clean public or local knowledge surface.
• Teams that want agents to retrieve from maintained docs instead of stale chat context.
• Researchers, builders, and PMs who want their thesis, project boundaries, and source trails to survive across AI sessions.

The reusable contract

your-wiki/
├── README.md
├── index.html
├── index.json
├── llms.txt
├── llms-full.txt
├── raw/<kb>/**/*.md
├── wiki/<kb>/**/*.html
└── scripts/pixi_wiki_mcp.py

Port your own knowledge base

• Create one folder per knowledge base.
• Add a README.md that explains scope, current status, and what is not covered.
• Add curated Markdown docs under stable paths.
• Generate raw Markdown mirrors and HTML pages.
• Generate index.json so tools can list KBs and documents.
• Generate llms.txt and llms-full.txt so agents have compact and full entrypoints.
• Run the read-only MCP server locally so agents can list, search, and read your KBs.

Copy this repo

The implementation lives in the public GitHub repo. Start there if you want to copy the shape and adapt it to your own use case.

Open the Pixi Wiki repo →

Recommended adaptation boundary

Publish only what you want public. Keep private voice, preferences, secrets, and sensitive strategy in a local-only KB or private repo, then connect that private KB to your agent client separately.