Content Model

Slingshot uses a shared content model across chat messages, community threads, and community replies. Every content entity carries the same field vocabulary so front-end clients can use a single rendering pipeline.

Content Layout

A content entity renders in this order:

1. Quote — if quotedMessageId / quotedReplyId is present, render the quotePreview block above the body.
2. Body — the main text, interpreted according to format.
3. Attachments — file/media previews from the attachments array.
4. Embeds — link previews from the embeds array.
5. Poll — if pollId is present, render the embedded poll widget.

format Field

Value	Behavior
'markdown' (default)	Parse body as GitHub-flavored Markdown, then parse inline tokens in text runs.
'plain'	Render body as literal text — no Markdown parsing. Inline tokens (<@userId>, <@&roleId>, etc.) are still parsed.

Note

The old contentType field (which included 'rich') has been replaced by format. The 'rich' option was removed — structured rich-text editors should serialize to Markdown.

Token Syntax

Tokens are inline markers embedded in the body text. They are parsed by parseContentTokens() from @lastshotlabs/slingshot-core/content.

Token	Syntax	Example	Purpose
User mention	<@userId>	<@user-123>	Mention a specific user.
Broadcast mention	<@everyone> or <@here>	<@everyone>	Mention all members or online members.
Role mention	<@&roleId>	<@&moderator>	Mention all users with a role.
Context reference	<#contextId>	<#channel-abc>	Link to a room/container/thread.
Emoji shortcode	:shortcode:	:wave:	Render an emoji by name.

Tokens inside code spans (`...`) and fenced code blocks are not parsed.

Field Reference

Chat Message

Field	Type	Notes
body	string	Required. Max 8,000 chars.
type	enum	'text' | 'image' | 'file' | 'gif' | 'sticker' | 'voice' | 'location' | 'contact' | 'system'
format	enum	'plain' | 'markdown' (default)
mentions	string[]	User IDs mentioned. Client-provided or parsed from body.
broadcastMentions	('everyone'|'here')[]	Broadcast mention tokens.
mentionedRoleIds	string[]	Role IDs mentioned.
attachments	AssetRef[]	Uploaded files/media. Max 10.
embeds	EmbedData[]	Server-resolved link previews.
quotedMessageId	string	ID of the quoted message (immutable).
quotePreview	QuotePreview	Snapshot of quoted message body + author.
pollId	string	Attached poll entity ID (immutable).
stickerId	string	Sticker asset ID (immutable).
location	LocationData	Geolocation for type: 'location'.
contact	ContactData	Contact card for type: 'contact'.
systemEvent	SystemEventData	Event payload for type: 'system'.
appMetadata	Record<string, unknown>	Opaque application metadata.

Community Thread

Field	Type	Notes
body	string?	Optional.
format	enum	'plain' | 'markdown' (default)
mentions	string[]	User IDs mentioned.
broadcastMentions	('everyone'|'here')[]	Broadcast mention tokens.
mentionedRoleIds	string[]	Role IDs mentioned.
attachments	AssetRef[]	Uploaded files/media.
embeds	EmbedData[]	Link previews.
pollId	string	Attached poll entity ID (immutable).

Community Reply

Field	Type	Notes
body	string	Required.
format	enum	'plain' | 'markdown' (default)
mentions	string[]	User IDs mentioned.
broadcastMentions	('everyone'|'here')[]	Broadcast mention tokens.
mentionedRoleIds	string[]	Role IDs mentioned.
attachments	AssetRef[]	Uploaded files/media.
embeds	EmbedData[]	Link previews.
quotedReplyId	string	ID of the quoted reply (immutable).
quotePreview	QuotePreview	Snapshot of quoted reply body + author.

Shared Types

All content-model types are exported from @lastshotlabs/slingshot-core:

import type {
  AssetRef,
  ContactData,
  ContentFormat,
  EmbedData,
  LocationData,
  QuotePreview,
  SystemEventData,
  VoiceMetadata,
} from '@lastshotlabs/slingshot-core';

Validation schemas (Zod) are on the /content subpath:

import {
  assetRefSchema,
  contactDataSchema,
  embedDataSchema,
  locationDataSchema,
  quotePreviewSchema,
  systemEventDataSchema,
} from '@lastshotlabs/slingshot-core/content';

Constants:

import {
  // 10
  MAX_CONTENT_ATTACHMENTS,
  // 8,000
  MAX_CONTENT_BODY_LENGTH,
  // 50
  MAX_CONTENT_MENTIONS,
} from '@lastshotlabs/slingshot-core';

Rendering Guide

Step 1: Parse tokens

import { parseContentTokens } from '@lastshotlabs/slingshot-core/content';

const message = { body: 'Hello <@user-123>' };
const parsed = parseContentTokens(message.body);
// parsed.segments — array of ContentSegment
// parsed.mentionedUserIds — string[]
// parsed.broadcastMentions — ('everyone' | 'here')[]

Step 2: Render segments

Walk the segments array. Each segment has a type discriminator:

Segment type	Rendering
text	Render as plain text (or Markdown if format === 'markdown').
mention	Render as a user mention chip. Look up display name by userId.
broadcastMention	Render as @everyone or @here badge.
roleMention	Render as a role mention chip.
contextRef	Render as a clickable link to the referenced room/container.
emoji	Render the shortcode as an emoji image or native character.
codeSpan	Render as inline code.
codeBlock	Render as a fenced code block with optional language tag.

Step 3: Append sidecar content

After the body, render in order: attachments → embeds → poll.

Search Indexing

When content is indexed by slingshot-search, the body is stripped of tokens via stripContentTokens() before indexing. This ensures that <@user-123> does not pollute full-text search results. The raw token data is preserved in the mentions, broadcastMentions, and mentionedRoleIds fields for filtered search.