Glossary
A quick, plain-language dictionary of the terms you'll meet across Altretta. Where a term has its own section, follow the link to learn it in depth.
Notes & writing
Vault — Your knowledge base: a folder of Markdown files on your own computer that Altretta reads and writes. You own it; there's no lock-in. See First run & your first vault.
Note — A single Markdown file in your vault — one idea, page, or document. See Your first note.
Wikilink — A link between notes written as [[Note name]]. Typing one connects two notes and builds your graph; its reverse shows up as a backlink.
Backlink — The reverse of a wikilink: the list of notes that link to the note you're reading. See Backlinks.
Unlinked mention — A place where a note's name appears as plain text but isn't yet a wikilink — a suggested connection you can turn into a real link. See Backlinks.
Property — A structured field on a note, stored in its frontmatter (e.g. status: draft, tags: [...]). Properties power Bases and filtering. See Tags & properties.
Tag — A #label keyword you attach to notes to group them across folders. See Tags & properties.
Canvas — An infinite, freeform board where you arrange notes, cards, images, and connections spatially — for mapping and thinking visually. See Canvas.
Base — A live table, card, or list view over a set of notes, driven by their properties — like a database view of your vault. See Bases.
Map & Manual — An auto-generated overview of an area of your vault: a map of how its notes connect plus a manual summarizing them, so you (and your AI) can orient quickly. Surfaced to the AI via aingle_vault_map. See Maps & Manuals.
Graph & structure
Graph — The visual network of your notes and the links between them — local (around one note) or global (your whole vault). See Graph view and the star-map.
Semantic graph — The real graph database underneath Altretta, storing your knowledge as triples (subject–predicate–object facts), not just wiki-links. It's what SPARQL and pattern queries run against. See The semantic graph and the MCP tool reference.
Symbol graph — For code: the graph of symbols (functions, types, files) and their cross-file relationships that Altretta builds when it maps a codebase. See Symbols.
Impact Radius — For code: the answer to "what breaks if I change this?" — the set of symbols and files affected by a change, computed from the symbol graph. See Impact Radius.
Verifiability
DAG (signed action history) — Altretta's append-only, hash-linked, Ed25519-signed log of changes to your vault (a directed acyclic graph). Knowledge that goes away is recorded as a signed retraction rather than vanishing from the record. This is what makes time-travel and provenance possible. See Signed history.
Provenance anchor — A cryptographic reference tying a retrieved passage to the exact state of your note in the signed history — proof of what the note said and when. Every grounded answer carries one. See Provenance.
Proof — A cryptographic object that confirms a fact matches the signed history. Fetched with aingle_get_proof and checked with aingle_verify_proof (see the MCP tool reference). See Provenance.
Groundedness — An honest verdict on how well the retrieved evidence supports an answer (e.g. grounded / weak / ungrounded). It's how Altretta admits when the notes don't actually say something. See Grounded, cited answers.
Citation (source:lines) — The way Altretta cites evidence: a note path plus a line range, e.g. decisions/2026-03-db.md:12-19. Every claim points back to its exact source. See Grounded, cited answers.
Time-travel — Browsing, diffing, and restoring any past version of a note from the signed history — because the DAG preserves everything. See Time-travel.
Signing key — Your personal Ed25519 key that signs every change you make, so the signed history can prove who recorded an action. See Your identity & signing key.
AI & connections
MCP — The Model Context Protocol, the open standard Altretta uses to expose your vault to any AI client over a small local endpoint. See Connect your AI and the MCP tool reference.
Review inbox — The place where AI-proposed changes wait for your approval before they're written back to the vault — nothing lands without consent. See The review inbox.
Publishing & the seal
Seal (証) — Altretta's mark. On a hosted published site whose notes carry a valid signature it reads "Verified · Altretta" and links to the site's verify URL; everywhere else it reads "Published with Altretta" and claims nothing. See The verify URL & Verified badge.
Teams & plans
Team brain — A shared, end-to-end encrypted vault for a team, with roles and cryptographic revocation, so knowledge is shared without giving up privacy or control. See Encrypted team brain.
Seat — A paid membership in a team or plan. Every member on a team vault's roster consumes one, whatever their role — Readers included. See Roles & permissions.
Reader — A read-only member of a team brain: they can read shared knowledge but not edit it. Like every other role, a Reader consumes a seat. See Roles & permissions.
Related
Reference overview
The section this glossary belongs to.
MCP tool reference
The tools these terms show up in.
Verifiable knowledge
The DAG, proofs, and time-travel in depth.
AI you can trust
Grounding, citations, and provenance in practice.
The semantic graph
The triple store behind the graph terms above.
Code Cartography
Where symbol graph and Impact Radius come from.