weave is an open-source, self-hosted alternative to Airtable, Fibery, Notion databases, and ClickUp, built so that AI agents are first-class users. Start at the project README — it covers what weave is, the quickstart, self-hosting, and the FAQ.
| Document | What it covers |
|---|---|
| Ontology & glossary | The one core kind — workspaces, spaces, tables and rows are all entities — how the levels nest, and the glossary; plus why an entity type is not a field type |
| Architecture map | Interactive map of the runtime: adapters, request core, engine, stores, renderers, agent surfaces; every node links to its code at a pinned commit |
| Feature parity matrix | weave scored against Fibery’s core work-platform feature set, feature by feature |
| weave vs Airtable | Concept mapping, field types, what wins where, how to migrate |
| weave vs Fibery | The closest model to weave’s, and the honest gaps |
| Open-source alternatives compared | weave against NocoDB, Baserow, Grist, and Teable |
| Screenshots | Table, entity documents, docked entity, relation map, search. Regenerate with node scripts/screenshots.mjs |
| Document | What it covers |
|---|---|
| AGENTS.md | Agent-facing map: repo layout, the 60 MCP tools, entity-ref forms, rules for changing the repo |
| CONTRIBUTING.md | Setup, the zero-dependency and no-build-step rules, PR expectations |
| SECURITY.md | Threat model, what is in scope, private reporting |
| llms.txt | Machine-readable index of this documentation |
Every running instance provisions a self-documenting weave workspace at
/w/weave/ — a Handbook, a Wiki, the roadmap and issue tracker, and a Quality
space mirroring the test suite. It is queryable through the same REST API as any
other workspace, so the tool documents itself in its own data model.