> For the complete documentation index, see [llms.txt](https://docs.katara.ai/about/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.katara.ai/about/reference/availability.md).

# Feature availability

These guides describe functionality found in the current Katara application source. They do not certify that your hosted environment has deployed every change or enabled every optional service. Permissions, subscription capabilities, configuration and resource readiness still apply.

## Current dashboard workflows

| Workflow                                        | Where to find it              | Important boundary                                                                         |
| ----------------------------------------------- | ----------------------------- | ------------------------------------------------------------------------------------------ |
| Passwordless sign-in and organization selection | Sign-in and organization menu | Use the invited account; active membership is required.                                    |
| Content Uploader and Q\&A creation              | Agent Library                 | These are the current normal agent-creation choices.                                       |
| Documents, filters, tags and collections        | Content Management            | Collections select documents; they do not attach them to a Brain.                          |
| Curated answers and spreadsheet import/export   | Q\&A Dashboard                | This is separate from Brain GitHub export.                                                 |
| Question testing                                | Prompt Lab                    | Requires an authorized Q\&A agent and generation permission.                               |
| Members, groups, grants and billing             | Organization menu             | Available actions depend on current authority.                                             |
| Audit events                                    | Audit                         | Visible only with the required capability; not a promise of a complete compliance archive. |
| OAuth MCP connection                            | `/mcp`                        | MCP User is read-only and does not gain the business dashboard.                            |

## Optional Brain capabilities

Brains have a dedicated creation and management experience. Knowledge, Sources, Activity and Settings are implemented, but the backend enables knowledge, Brain-backed answering and export separately. The configuration switches `KNOWLEDGE_ENGINE_ENABLED`, `KNOWLEDGE_ENGINE_QUERY_ENABLED` and `KNOWLEDGE_ENGINE_EXPORT_ENABLED` default to false in the inspected source.

If your deployment does not enable one of these services, a visible UI surface does not make the operation available. Ask your administrator about rollout status rather than trying different credentials or creating duplicate resources.

Newer contextual Wiki editing and Brain settings/ownership workflows were inspected in working trees with uncommitted application changes. They are documented as current-source functionality, not confirmed released features. Before relying on them operationally, confirm the frontend, published client and backend have been deployed together.

The Knowledge landing does not offer a **New page** workflow. The current Wiki reader can offer **Edit page** for an existing readable page to an authorized contributor. MCP/API page operations are a separate surface; do not infer a dashboard button from their existence.

## Backend features without a current creation journey

Taxonomy and the Flesch, Language, Topic, Passive Voice, Sentiment, Diátaxis, Tone and General LLM classifiers remain in the inspected backend's accepted/runnable agent registry. They are **not current dashboard creation choices**, and the frontend may treat those records as unsupported types.

This site therefore does not teach a classifier-creation UI that does not exist. If you maintain an API-based workflow, confirm its deployed API contract, entitlement and operational support with your administrator. The backend/frontend mismatch is not evidence that classifiers have been retired from the backend.

## Retired and retained integrations

* **Website, GitHub and Notion ingestion agents:** new execution is retired in the current backend. Historical records and some frontend forms may remain; those forms do not establish a working ingestion service.
* **GitHub Brain export:** a different, optional capability that publishes committed Wiki history. It is not a replacement name for GitHub ingestion or bidirectional repository sync.
* **Discord, Telegram, Slack and Website Widget:** retained existing-agent surfaces are present, but the current creation menu does not offer new instances. External installation and credentials need separate verification.
* **Google Drive:** a direct integration route exists, but it is absent from the main sidebar and its persistent queue/callback work is incomplete. It is not a standard onboarding path.
* **Workflows index:** older route parameters and components do not constitute a current Agent Library Workflows tab. Some existing-agent workflow views remain.
* **Real-time question type:** the current Q\&A form treats it like Blocked; do not expect a bypass-cache generation mode.
* **EU residency:** marked Coming soon in current onboarding.

See [Existing integrations](https://docs.katara.ai/about/agents-and-answering/integrations.md) for maintenance guidance.

## MCP and API boundaries

The MCP server has a verbs surface and an extended surface; actual discovery depends on deployment and credential scopes. See [Tools and source selection](https://docs.katara.ai/about/connect-ai-clients/tools.md).

Collections are implemented as Synchronizer agents. The dedicated collection `ListCollections` and `RefreshCollection` handlers return Unimplemented and direct callers to AgentService; they are not working alternatives for an integration.

The application does not currently provide the generic **AI Gateway** or third-party **MCP Registry** administration journeys described in older versions of this documentation. Do not use that old terminology as a setup guide for the current app.
