EonityGitHub

Semantic Asset Platform · Open source · early access

Control your
knowledge flow.

TYDAL is the typed Digital Asset Layer. Turn documents, images, datasets and code into structured, searchable resources. Use schemas to define their meaning and Vaults to make selected content available to people, applications and AI agents.

Built first for photography collections and museums, and for AI knowledge work. Archives and learning resources next.

Get started View on GitHub

Run your own instance with Docker, or connect an AI client to one you already use.

Your content. Your structure. Your choice of AI.

Your knowledge flowRESOURCES → VAULTS
TYDAL repositoryYour resources, structured and enriched
CORE
DocumentsText & knowledge
ImagesMedia & meaning
DataStructure & context
SchemasAI enrichmentSearch
VaultsContextual views of your resources
Selected resourcesContextAccess rules

Publish selected resources as galleries and knowledge spaces.

One repository. Many experiences. Concept illustration
Open sourceSchema-drivenAPI & MCPProvider-flexible AI

Beyond digital asset management

From digital assets
to semantic knowledge.

Typed means your resources follow schemas: fields, validation and search behavior are defined together. Find content by meaning, connect related resources and reuse them across different contexts.

The Photo Exhibition scheme in TYDAL: fields for title, description, author, year, technique and dimensions, each with its storage, type, facet and whether it is required.01

Organize.

Keep an asset, its supporting files and representations together as a resource. Collection schemas define its metadata, file roles and indexing rules.

Schema-driven collections
Two photographs in the TYDAL library, each with an AI-written title such as “Windsock and Skydiver Under Clear Blue Sky” and typed tags like skydiver, parachute and sky.02

Understand.

AI can suggest descriptions, tags and classifications from your content. Review those suggestions before applying them, or use configurable approval rules.

AI enrichment with review
The Vaults list in TYDAL: three gallery Vaults, City Lines, Pike and Shot and Scale, from two organizations, one private and two public.03

Explore & publish.

Publish selected resources through Vaults with their own context and access rules. Reuse the same source across galleries, applications and AI connections.

One source, multiple contexts
TYDAL repositoryContextual VaultsExperiencesMeet the model

Built around TYDAL

One foundation.
Many ways to build.

TYDAL provides the shared resource platform; each client adds its own experience. Documents, images, datasets and code can serve different applications through Vaults, with their own context and access rules.

01FullFrame

A photography client.
Built on TYDAL.

TYDAL's first standalone open-source client adds exhibitions, curator selection and an optional jury. Photographs remain in TYDAL, available to other applications through their Vaults.

Photography · Museums · ArchivesThe FullFrame home page: “Make room for a different view.” beside the first two of its four exhibition cards, Scale and Pike and Shot, each from a different organization.
See how the client connects FullFrame on GitHub
02AI connections

Bring your AI
to your knowledge.

Connect external AI tools through MCP to explore the resources exposed by a Vault, including their content, metadata and relationships.

Agents & knowledge workflowsClaude reports that all six photographs in Drift have new titles and descriptions, written through the organization’s MCP server and checked in the Drift Vault, with a table of the new titles from “Last Breath of Air” to “Which Way the Wind Blows”.
Connect an AI client
03Your application

Make something
of your own.

Build an experience around your content using the API and SDK—from a resource library to a learning application.

API · TypeScript SDK · VaultsA Vault’s capabilities in TYDAL: its purpose preset and read tiers that decide whether text, files and questions are answered, each with its own setting.
Build your own client

See it work

Your AI proposes.
You decide.

Connected to TYDAL through MCP, Claude looks at each photograph in Drift, proposes titles and descriptions, and asks before it writes anything back.

Claude reads the exhibition through its vault and writes the titles back through the organization’s MCP server: the vault decides what the AI sees, the organization what it may change.
FullFrame’s Wall view of Drift: a parachutist lands under a blue and white canopy, titled “Last Breath of Air”, with the description Claude wrote below it.
The result, live in FullFrame. Nothing was exported or uploaded again: FullFrame reads the new titles from TYDAL.

For the people who build

The semantic layer
between your files
and your AI.

Manage resources and collections through the repository API. Build applications against Vault interfaces, or connect AI agents through MCP. Each Vault controls the content and capabilities it exposes.

Explore developer docs
Read your Vault metadata
# Use your instance’s Vault URL and key
curl "$TYDAL_VAULT_URL/meta" \
  -H "X-Vault-Key: $TYDAL_VAULT_KEY"
HTTP APIVault-scoped access

Built in the open

Good knowledge
deserves open tools.

TYDAL is built for developers and early adopters who want to understand, adapt and extend the tools they use.

TYDAL is available under the Apache License 2.0. Run your own instance, inspect the code and contribute to its development.

Explore the repository

Read the contribution guide

Clients can be open source too. FullFrame is available under Apache-2.0: explore a working application built on TYDAL's Vault interfaces.