weave: portable work system for people and their agents

weave and Fibery

weave is a portable work system you own, developed in the open. Fibery is a managed workspace for connected databases, collaborative documents and visual planning. Both organize related records and expose programmatic interfaces. Their differences concern ownership, operating model and how much work the application handles directly on screen.

Reviewed November 20, 2026 against weave v0.4.165 source and the linked Fibery documentation. This is a capability review, not a comparative performance test. The historical feature checklist yields 85.5% under its own equal-weight scoring rule. That number does not establish current parity, compatibility, reliability or fit.

Capability comparison

Area weave Fibery
Ownership MIT source; local, self-hosted or weave cloud Proprietary managed service
Agent workflow Native MCP, REST and CLI; use the agent you already have Built-in AI and external MCP access
Record structure Spaces, tables and fields exposed as system records Connected databases configured through visual and programmatic interfaces
Relations Ordinary bidirectional relations; multi-target relations can include structure records Connected database relations; check current documentation for polymorphic support
Documents Markdown fields on records, addressable as Markdown, HTML or PDF Collaborative rich text on entities and standalone documents
Planning Table, list and board layouts, gallery-style cards and relation map Also timelines, calendars, whiteboards and reports
Automation Native polling and inbound hooks, schedules, HTTP actions, persisted outbound retries Rules, buttons, scripts and managed integrations
Capture Forms with stable keys, URL drafts, HTTP/CLI/MCP submission and retry receipts Forms and URL prefilling
Custom interfaces HTML documents, API clients and modification of the source Supported Custom Apps, including external-agent development
Permissions Workspace and space roles, private spaces, scoped tokens, share links and credential grants Richer sharing and access templates for differentiated team access
Search Text retrieval; semantic retrieval remains proposed Full-text and semantic search
Recovery SQLite storage, JSON interchange, backup/restore, undo and document revisions Managed infrastructure and in-app history
Development Public source and tests; roadmap and issues in weave Public roadmap, changelog and community; proprietary implementation

weave’s fit

You already have an agent. weave works natively with it. Its engine serves the browser, MCP, REST and CLI, so an existing agent can operate records, documents, structure and automation. Fibery also supports external agents; MCP itself is not an exclusive capability.

A document has a structured home in a record. Multi-target relations extend that model to different allowed record types, including tables and spaces. They do not have inverse fields, and lookups and rollups do not support them. Relationships with properties still need a junction record. See polymorphic relations.

Native listening connects external changes to the same record system, with visible rules and failure recovery. The first poll establishes a baseline and missing source items do not generate deletion events. Incoming webhooks and polling are established patterns, not unique to weave. See automations.

Ownership extends to the runtime. A workspace uses SQLite storage, with attachments and secrets handled separately. Source, backup and restore let an operator run it independently. The small runtime does not prove faster queries or greater scale. HTML documents offer programmable content, but their scripts execute on the application origin and require trusted authors.

Fibery’s fit

Fibery supports more collaborative authoring and visual planning within the application. Timelines, reports and whiteboards matter when the team performs that work directly on screen. Its Custom Apps provide a supported route to custom interfaces, and its managed integrations reduce operator work.

Standalone documents are optional: a team can use document fields on entities in either product. Avoid treating sidebar clutter as an unavoidable Fibery property. Access design also deserves a workflow-specific review; credential grants in weave are not ordinary field-level permissions.

Choosing and migrating

Start with one representative workflow. Test its relations, document formats, formula outputs, attachments, access boundaries and automation failures. A CSV export does not recreate permissions, formulas, documents or automation. Rebuild and verify those deliberately before moving the authoritative workflow.

The product principles describe weave on its own terms, including the maintenance commitment and proposed capabilities. The README matrix summarizes current choices.

Sources