Axiom Wiki

What is Axiom Wiki?

Unlike RAG systems that re-derive answers from raw sources on every query, Axiom compiles knowledge into an interconnected wiki of markdown pages and keeps it current as new sources arrive.

Drop files, get a wiki

Drop PDFs, markdown, images, spreadsheets (CSV/TSV), or URLs into your raw/ folder. Axiom reads them, extracts entities and concepts, and creates cross-linked wiki pages automatically.

Auto-wiki your codebase

Run axiom-wiki autowiki inside any project to generate structured documentation — architecture, modules, tech stack — with one command.

Stay in sync

As your codebase evolves, axiom-wiki sync detects changes and updates only the affected wiki pages. No manual maintenance.

Hybrid Semantic Search

Search that actually understands. Combines traditional keywords with vector embeddings for conceptual retrieval. Supports Google, OpenAI, and Ollama.

Live Maintenance

The wiki updates itself when your code changes. Tier 1 (deterministic) handles renames; Tier 2 (agent-based) rewrites pages to reflect new logic.

Forensic Citation Proofs

Zero hallucinated citations. Mechanically audit paragraph citations against raw files — timestamps ([mm:ss]), pages ([p. N]), and quotes. Run axiom-wiki lint --forensic in CI or git pre-commit hooks.

Strict Grounded Querying

Query your wiki with zero extrapolation. In strict forensic mode (axiom-wiki query --forensic), the model is strictly barred from pre-trained knowledge and only speaks from verified docs.

Browse it like a real wiki

axiom-wiki serve opens a local read-only web UI with a dashboard, hybrid search, a link graph, and “what links here” backlinks on every page.

Works everywhere

Nine providers from Google Gemini’s free tier to Grok, plus fully offline with Ollama. Your data stays local. Plain markdown files work with Obsidian, VS Code, or any editor.