> ## Documentation Index
> Fetch the complete documentation index at: https://replyke-feat-push-rich-payload-fields.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Reputation

> Per-space reputation buckets, the maintained overall total, and the opt-in read contract for space-scoped reputation

Reputation in Sublay is **space-scoped**. Instead of a single number per user, each user accrues reputation into separate **buckets** — one per space, plus a project-general bucket — and their `reputation` field is the maintained sum of all of them.

Reputation is owned by the optional **`reputation`** bundle. Install it from the **Database** page of your project to turn the feature on. The `reputation` bundle works with or without the `spaces` bundle: without spaces, only the general bucket accrues. See [Bundles](/bundles).

## Buckets

The bundle provisions a per-tenant table of reputation buckets, keyed by `(userId, spaceId)`.

| Bucket          | Key                   | Meaning                                           |
| --------------- | --------------------- | ------------------------------------------------- |
| Project-general | `spaceId IS NULL`     | The user's reputation outside of any space.       |
| Per-space       | `spaceId = <spaceId>` | The user's reputation within that specific space. |

A few rules govern buckets:

* **Lazy creation** — a bucket is created the first time a user's reputation changes for that `(user, space)` pair. A bucket that has never been touched reads as `0`.
* **Negative scores allowed** — there is no floor; a bucket can go below zero.
* **Buckets are not returned directly** — there is no endpoint that lists raw buckets. You read reputation through the `reputation` total and the opt-in `spaceReputation` field described below.

## The maintained total

Every user object still carries a `reputation` field, and default API responses are unchanged. `reputation` is the **maintained overall total** — the sum of all of that user's buckets (general plus every space). It is always present, whether or not a space is in context.

## Reading space-scoped reputation

Every user-returning endpoint accepts an opt-in `spaceReputationId` query param that adds a `spaceReputation` number to each returned user. The `reputation` total is always included regardless; `spaceReputation` is purely additive and only appears when you ask for it.

<ParamField query="spaceReputationId" type="string">
  Adds a `spaceReputation` field to each returned user. One of:

  * a space **`<uuid>`** — that user's reputation in that specific space;
  * **`none`** — the project-general (space-less) bucket;
  * **`context`** — the space derived from the request context. On lists this is **per-row** (each author scored in the space *that row* belongs to); on single-space surfaces — space members/team, a chat conversation — it's that one space. Only valid where a space is in context, so it is **rejected (400) on bare user lookups**.

  A bucket that doesn't exist reads as `0`. The empty string and the legacy `general` / `null` aliases are **rejected (400)** — use `none` for the space-less bucket.
</ParamField>

<ParamField query="spaceReputationDescendants" type="boolean">
  Only honored alongside an explicit space `<uuid>`. When `true`, `spaceReputation` is the **subtree sum** — the named space plus all of its descendant spaces (the root space's own bucket is included). Ignored for `none`; not allowed with `context`.
</ParamField>

The resulting field appears on each user object as:

<ResponseField name="spaceReputation" type="number">
  The user's reputation in the requested space (or the space-less bucket). Present only when `spaceReputationId` was supplied. Missing buckets read as `0`.
</ResponseField>

### Where the params apply

`spaceReputationId` is accepted on **every** user-returning endpoint, but the allowed values depend on whether the endpoint has a space in context.

**Context endpoints** accept `<uuid>`, `none`, **and** `context`:

* Entity feed and single entity
* Comment feed and single comment
* Entity and comment reaction lists
* Chat — list messages, get message, send message, list members, list reactions
* Space members and space team
* Moderated reports
* Content search and AI ask

**User-direct endpoints** accept `<uuid>` and `none` only — `context` is **rejected (400)** because there is no space context to derive:

* Fetch user (by ID, foreign ID, or username) and user suggestions
* Followers / following lists (by user ID)
* Connections list (by user ID)

Project-wide user search (`search-users`) and the self-relative follows/connections lists are **not** enriched and ignore the param.

## Uninstalling is lossy

Removing the `reputation` bundle drops the buckets table **and** zeroes the `reputation` column on every user. Reinstalling starts fresh from zero — past scores are not recomputed. Back up anything you need before uninstalling. See [Bundles](/bundles).

## See Also

* [Bundles](/bundles)
* [User data model](/data-models/user)
* [Space data model](/data-models/space)