Collections

A collection is a named set of related fragments within a workspace. Collections let you curate knowledge for a reader journey, a workflow, or a specific purpose without moving or duplicating the fragments.

A fragment can belong to more than one collection. Removing it from a collection does not delete the fragment; it only removes that membership.

What collections are for

Use a collection when a group of fragments should be presented or retrieved together. Common examples include:

• a getting-started path;
• authentication troubleshooting;
• support onboarding;
• documentation for a particular release;
• the source set used by a focused AI workflow.

The collection gives the group a name and description. The individual fragments keep their own titles, types, tags, and content.

Collections are workspace-scoped

A collection belongs to one workspace and contains fragments from that workspace. This keeps its membership within the same access and knowledge boundary.

If information must be made available across workspaces, use an intentional cross-workspace mechanism such as a fragment symlink, subject to permissions. A collection does not bypass workspace access.

A fragment can appear in several collections

Collections use many-to-many membership. One fragment can contribute to several curated experiences without being copied.

For example, “Configure an API token” might belong to:

• “Getting started for developers”;
• “API integration”; and
• “Authentication troubleshooting”.

Updating the fragment improves all three collections because they refer to the same source.

Collection, type, tag, or workspace?

These concepts organize knowledge at different levels:

• A workspace defines the knowledge and access boundary.
• A fragment type says what a fragment is, such as Knowledge or Recipe.
• A tag labels an attribute that may cut across many topics.
• A collection curates a named group of fragments.

Suppose you are documenting authentication. Keep it in the product documentation workspace. Use Knowledge, Recipe, and Solution types according to each fragment’s purpose. Apply the authentication tag where relevant. Create an “Authentication” collection only if readers or tools benefit from treating a selected set as one guide.

Names and descriptions

A collection name should tell readers what the set is for. “API onboarding” is clearer than “API docs” if the collection specifically guides new developers.

Use the description to explain the audience, scope, and expected outcome. This is especially important when similar collections exist or when AI tools can select a collection as context.

Collection names are intended to be distinct within their workspace. Avoid names that differ only by capitalization or punctuation.

Managing membership

Fragments can be assigned to collections while they are created or updated. Collection-focused interfaces and APIs can also add or remove one or more fragments.

Before bulk-adding content, review whether each fragment contributes to the collection’s purpose. A collection is useful because it is curated; automatically including every fragment with a broad tag may recreate the noise the collection was meant to reduce.

Deleting a collection removes the grouping, not the underlying knowledge. Verify your workflow before deletion if an application or AI client refers to the collection identifier.

Example: a getting-started collection

A “Getting started with Usable” collection might contain:

1. What Usable is
2. Create a workspace
3. Create your first fragment
4. Find knowledge with search
5. Connect Usable Chat or MCP

The fragments can still appear in other collections and search results. The collection simply defines a recommended path for new users.

Collections and search

Supported search and listing workflows can filter by collection or return collection membership with fragment results. Use this when the user has already chosen a curated scope.

For open-ended questions, semantic or agentic search across the workspace may be more appropriate. A collection is a deliberate subset, so it may exclude a relevant fragment that was never added.

Good practices

• Name collections for an audience, outcome, or stable scope.
• Write a description that defines what belongs.
• Keep membership curated rather than merely large.
• Reuse fragments across collections instead of duplicating content.
• Review release- or project-specific collections when their scope expires.
• Do not use collections to imitate access controls; permissions come from the workspace.

Related concepts

• Workspaces define the boundary in which a collection exists.
• Fragment types describe what each member fragment is.
• Tags and summaries support reusable labels and discovery.
• Search and retrieval explains when to search broadly or filter by a curated set.