Skip to content

Wiki Structure

my-project/
axiom/ # Everything lives here
config.json # Local config (provider, model, paths, obsidianCompat)
state.json # Compilation state (SHA-256 hashes, concepts, frozenSlugs)
map-state.json # Autowiki/sync state (pages, git hash)
search.index # Orama search index (binary/JSON)
raw/ # Source files to ingest
.axiomignore
assets/
wiki/
pages/
entities/ # People, places, organisations
concepts/ # Ideas, topics, theories
sources/ # One summary per source file
analyses/ # Filed answers, comparisons
index.md # Page catalog (by category)
moc.md # Map of Content (by tag)
log.md # Operation history
usage.log # Token usage and cost
schema.md # Wiki conventions
~/axiom/
state.json # Compilation state (SHA-256 hashes, concepts, frozenSlugs)
map-state.json # Autowiki/sync state
search.index # Orama search index
raw/ # Source files to ingest
.axiomignore
assets/
wiki/
pages/
entities/
concepts/
sources/
analyses/
index.md
moc.md
log.md
usage.log
schema.md

Global config lives in your OS config directory (~/.config/axiom-wiki/ on macOS/Linux).

Every wiki page uses this YAML frontmatter:

---
title: "Alan Turing"
summary: "British mathematician and pioneer of computer science"
tags: [mathematics, computing, ai]
category: entities
sources: ["turing-biography.pdf"]
updatedAt: "2026-04-10"
---
CategoryContains
entitiesPeople, places, organisations, named things
conceptsIdeas, topics, themes, theories
sourcesOne summary page per raw source file
analysesFiled answers, comparisons, syntheses

Internal links use wiki-link syntax:

[[entities/alan-turing]]
[[concepts/turing-completeness]]

With obsidianCompat: true in config, links use bare names instead:

[[alan-turing]]
[[turing-completeness]]

Axiom’s graph parser handles both formats — bare names default to the entities/ category.

Wiki pages use paragraph-level citations to trace information back to source files:

Alan Turing was a British mathematician who made foundational contributions
to computer science and artificial intelligence. ^[turing-biography.pdf]
His work at Bletchley Park was instrumental in breaking the Enigma code. ^[turing-biography.pdf] ^[intelligence-trap.md]

Every factual paragraph cites the source file(s) it was derived from.

The moc.md file is an auto-generated tag-grouped index. While index.md organizes pages by category, moc.md groups them by tag — each page appears under every tag it has:

## machine-learning
- [[pages/concepts/gradient-descent]] — Gradient Descent
- [[pages/concepts/neural-networks]] — Neural Networks
## history
- [[pages/entities/alan-turing]] — Alan Turing

It is rebuilt automatically after every ingest, watch, clip, and autowiki operation.

The index.md and log.md files are plain text — parseable with standard Unix tools:

Terminal window
grep "^## \[" wiki/log.md | tail -5 # last 5 operations
grep "ingest" wiki/log.md | wc -l # total sources ingested
grep "ingest" wiki/usage.log # cost breakdown per ingest