Integrations

The Signal Bot integrates with multiple external services to provide enhanced functionality. This page provides an overview of all integrations and their configuration.

Integration Status Overview

Service	Purpose	Commands	Status Check
PeerTube	Video hosting	`!save`	`./deploy.sh test-api`
Discourse	Forum posting	`!fpost`, `!flatest`, `!fsearch`	Check logs
Outline	Wiki docs	`!ol`, `!outline`	Check logs
Authentik	SSO accounts	`!createuser`, `!accountinvite`	Check logs
Local AI	Self-hosted AI	`!lai`, `!uai`	`./deploy.sh test-api`
OpenAI	AI assistant	`!ai`, `!summarize`	Check logs
MinIO	File storage	`!archive`, `!files`	`./deploy.sh test-minio`
Shlink	URL shortening	(auto)	`./deploy.sh test-shlink`
GitLab Wiki	Wiki content push	`!wikiadd`	Check logs
Gitea	Git repos	(auto 3D file archiving)	Check logs
Wikipedia	General knowledge	`!ask` (via unified search)	Public API
Google Knowledge Graph	Entity facts	`!ask` (via unified search)	Optional API key

PeerTube

Purpose: Upload and host videos from Signal

Commands:

!save - Upload video with AI-generated metadata

Configuration

PEERTUBE_URL=http://peertube:9000           # Internal Docker URL
PEERTUBE_PUBLIC_URL=https://videos.irregularchat.com
PEERTUBE_ACCESS_TOKEN=<long-lived-token>
PEERTUBE_CHANNEL_ID=1

Getting an Access Token

Log into PeerTube as an admin
Go to Settings > Application tokens
Create a token with upload permissions
Set expiry to far future (e.g., 2030)

How It Works

User replies to video with !save
Bot downloads video from Signal
AI analyzes frames + audio transcript for metadata
Video uploaded to PeerTube with generated title/description
Bot returns watch link (SSO login required)

Troubleshooting

# Test connectivity
./deploy.sh test-api

# Check PeerTube is reachable
docker exec signal-bot-selfhosted-signal-bot-1 curl -s http://peertube:9000/api/v1/config

Discourse

Purpose: Post content to community forum

Commands:

!fpost - Create forum post
!flatest - Show latest posts
!fsearch <query> - Search forum
!categories - List categories

Auto-features:

News articles auto-posted with AI summaries
Breakout room summaries posted to forum

Configuration

DISCOURSE_URL=https://forum.irregularchat.com
DISCOURSE_API_KEY=<api-key>
DISCOURSE_USERNAME=bot.username
DISCOURSE_QA_CATEGORY=5    # Category ID for Q&A/news posts

Getting API Key

Go to Admin > API > API Keys
Create a key for the bot user
Set appropriate permissions (create topics, post)

Getting Category ID

# Via API
curl -H "Api-Key: <key>" -H "Api-Username: <user>" \
  "https://forum.irregularchat.com/categories.json" | jq '.category_list.categories[] | {id, name}'

# Or check URL: /c/category-name/5 (5 is the ID)

Outline

Purpose: Create and search team documentation

Commands:

!ol <query> - Search docs
!ol ask <question> - AI Q&A from docs
!ol note <text> - Quick note
!ol ai <content> - Create doc with AI structure
!ol -n 5 - Capture last 5 messages to doc

Configuration

OUTLINE_URL=https://outline.irregularchat.com
OUTLINE_API_KEY=<api-key>
OUTLINE_DEFAULT_COLLECTION=<collection-uuid>

Getting API Key

Log into Outline
Go to Settings > API
Create a new API key

Breakout Room Integration

When a breakout room ends:

AI generates summary with decisions/action items
Summary posted to Outline
Link shared in original channel

Authentik SSO

Purpose: Create SSO accounts for new members

Commands:

!createuser @user email - Create account
!accountinvite [hours] - Create invite link

Configuration

AUTHENTIK_BASE_URL=https://sso.irregularchat.com
AUTHENTIK_API_TOKEN=<api-token>
AUTHENTIK_INVITE_FLOW_ID=<flow-uuid>
AUTHENTIK_INVITE_FLOW_SLUG=invite-enrollment-flow

Getting API Token

Go to Admin > System > Tokens
Create a new token with appropriate permissions
Token needs: user creation, flow access

Getting Flow ID

curl -s -H "Authorization: Bearer <token>" \
  "https://sso.irregularchat.com/api/v3/flows/instances/" \
  | jq '.results[] | "\(.pk) - \(.slug) - \(.name)"'

Look for invite-enrollment-flow or similar enrollment flow.

Local AI (Open WebUI)

Purpose: Self-hosted AI for unrestricted queries

Commands:

!lai <question> - Community AI model
!uai <question> - Uncensored model

Configuration

LOCAL_AI_URL=https://ai.irregularchat.com
LOCAL_AI_API_KEY=<api-key>
LOCAL_AI_MODEL=irregularbot

Model Selection

Command	Model	Behavior
`!lai`	irregularbot	Moderate filtering
`!uai`	uncensored	No refusals

Test Connectivity

./deploy.sh test-api

OpenAI

Purpose: AI assistant, summarization, embeddings

Commands:

!ask <question> - Unified agentic search (router, gap detection, synthesis)
!ai <question> - GPT-4o-mini questions
!summarize <url> - Summarize content
!tldr <url> - Quick summary
!wikiask <question> - Wiki Q&A (uses embeddings)

Auto-features:

Unified search routing, gap detection, and answer synthesis
News article summarization
Video metadata generation
Breakout room summaries

Configuration

OPENAI_API_KEY=sk-...

Rate Limiting

20 requests per hour per user
Group chats only (security)

MinIO S3

Purpose: File storage and archiving

Commands:

!archive - Archive a file
!files <query> - Search archive
!get <file> - Get file link

Auto-features:

Auto-scan and archive attachments
Meme storage

Configuration

MINIO_ENDPOINT=http://minio-irregularchat:9000    # Internal
MINIO_ACCESS_KEY=<access-key>
MINIO_SECRET_KEY=<secret-key>
MINIO_BUCKET=irregularchat
MINIO_REGION=us-east-1
MINIO_PUBLIC_URL=https://s3.irregular.chat        # Public for presigned URLs

Dual Client Architecture

MinIO requires TWO S3 clients:

Internal Client (MINIO_ENDPOINT): For API operations
Presign Client (MINIO_PUBLIC_URL): For generating presigned URLs

This ensures signature verification works correctly.

Test Connection

./deploy.sh test-minio

Shlink URL Shortener

Purpose: Shorten long presigned URLs

Auto-feature: Automatically shortens MinIO presigned URLs for cleaner display

Configuration

SHLINK_API_URL=http://shlink-api:8080
SHLINK_API_KEY=<api-key>
SHLINK_PUBLIC_URL=https://irregular.chat

How It Works

MinIO generates presigned URL (long, contains signature)
Shlink creates short URL pointing to presigned URL
Short URL expires when presigned URL expires

Test Connection

./deploy.sh test-shlink

Gitea

Purpose: Archive 3D printing files to git

Auto-feature: When 3D file links (Thingiverse, Cults3D, Printables) are shared, files are archived to Gitea

Configuration

GITEA_API_URL=https://git.irregularchat.com/api/v1
GITEA_TOKEN=<api-token>
GITEA_DEFAULT_OWNER=irregulars
GITEA_DEFAULT_REPO=3d-files

Supported Sites

Thingiverse
Cults3D
Printables
MyMiniFactory

GitLab Wiki (IrregularPedia)

Purpose: Push AI-extracted content from Signal chats to the community wiki

Commands:

!wikiadd - Reply to a message to extract and add wiki-worthy content
!wikiadd -n <count> - Analyze last N messages for facts and resources
!wa -n <count> - Alias for !wikiadd -n

Configuration

GITLAB_WIKI_URL=http://gitlab:8929/irregulars/IrregularChatWiki.git   # Internal Docker URL
GITLAB_WIKI_TOKEN=<personal-access-token>                              # write_repository scope
WIKI_WORKDIR_PATH=/app/wiki-workdir                                    # Git clone directory
WIKI_GIT_USER=IrregularBot                                             # Commit author name
WIKI_GIT_EMAIL=bot@irregularchat.com                                   # Commit author email

Setup Requirements

GitLab bot user with Maintainer access on the wiki project (required to push to protected main branch)
Personal access token with write_repository scope
Docker volume for wiki-workdir (persists git clone across container restarts)
Docker network — bot and GitLab must share a network (e.g., postgres-network) so the bot can reach gitlab:8929

How It Works

User runs !wikiadd (reply or -n count)
Bot extracts messages from Signal chat history
OpenAI analyzes messages for factual, educational content
Bot searches existing wiki pages by title AND content snippets
AI decides: append to existing page or create a new page
Bot clones/pulls the wiki repo into wiki-workdir
Changes are committed and pushed to GitLab
Astro Starlight rebuilds the site (irregularpedia.org)

Security

Git credentials are embedded in the remote URL at runtime, never logged
All error messages are sanitized to prevent token leakage
File paths are validated against path-traversal attacks
YAML frontmatter and Markdown content are sanitized against injection
A mutex lock prevents concurrent wiki operations

Troubleshooting

# Check wiki-workdir permissions (must be writable by node user, UID 1000)
docker exec signal-bot-selfhosted-signal-bot-1 ls -la /app/wiki-workdir

# Test GitLab connectivity from inside container
docker exec signal-bot-selfhosted-signal-bot-1 git ls-remote http://gitlab:8929/irregulars/IrregularChatWiki.git

# Check bot user access level (must be 40 = Maintainer)
# Via GitLab rails console on the GitLab container

Key Files

src/src/utils/wiki-content-manager.ts — Core module (clone, analyze, commit, push)
src/src/bot/command-handler.ts — handleWikiAdd() method

Wikipedia

Purpose: General knowledge search for the !ask unified search command

Used by: !ask (unified search) — selected by the AI router for general knowledge, concepts, people, organizations, and history questions.

How It Works

AI router determines the question needs general knowledge
Wikipedia MediaWiki Action API is queried with the question
Top articles are fetched with summaries (extracts)
Results are included in the unified search context for synthesis

Configuration

No API key or authentication required. Uses the public MediaWiki Action API.

Rate limit: 200 requests/second (well above bot usage)

Key File

src/src/utils/wikipedia-client.ts

Google Knowledge Graph

Purpose: Quick structured facts about well-known entities (people, organizations, places, concepts)

Used by: !ask (unified search) — selected by the AI router for entity disambiguation and quick facts.

How It Works

AI router determines the question involves a well-known entity
Google Knowledge Graph Search API returns structured data
Entity descriptions and detailed summaries are extracted
Results are included in the unified search context

Configuration

GOOGLE_KG_API_KEY=<your-api-key>    # Optional — if not set, Knowledge Graph is silently skipped

Getting an API Key

Go to Google Cloud Console
Enable the “Knowledge Graph Search API”
Create an API key (no OAuth required)
Free tier: 100,000 queries/day

Key File

src/src/utils/knowledge-graph-client.ts

External APIs (Optional)

Thingiverse

For 3D file metadata extraction:

THINGIVERSE_ACCESS_TOKEN=<token>

Cults3D

CULTS3D_API_KEY=<api-key>

Research Tools

For !research command:

RESEARCHTOOLS_API_URL=<url>

Integration Testing

Test All Integrations

# PeerTube + Local AI
./deploy.sh test-api

# MinIO S3
./deploy.sh test-minio

# Shlink
./deploy.sh test-shlink

# Check logs for other integrations
./deploy.sh logs -t 100 | grep -E "(Authentik|Outline|Discourse)"

Verify in Container

./deploy.sh shell

# Inside container:
env | grep -E "(PEERTUBE|DISCOURSE|OUTLINE|AUTHENTIK)"

Configuration Reference - All environment variables
Automated Features - Background processes
File Management - Archive and file commands
Deployment Guide - Server setup

Integrations

Integration Status Overview

PeerTube

Configuration

Getting an Access Token

How It Works

Troubleshooting

Discourse

Configuration

Getting API Key

Getting Category ID

Outline

Configuration

Getting API Key

Breakout Room Integration

Authentik SSO

Configuration

Getting API Token

Getting Flow ID

Local AI (Open WebUI)

Configuration

Model Selection

Test Connectivity

OpenAI

Configuration

Rate Limiting

MinIO S3

Configuration

Dual Client Architecture

Test Connection

Shlink URL Shortener

Configuration

How It Works

Test Connection

Gitea

Configuration

Supported Sites

GitLab Wiki (IrregularPedia)

Configuration

Setup Requirements

How It Works

Security

Troubleshooting

Key Files

Wikipedia

How It Works

Configuration

Key File

Google Knowledge Graph

How It Works

Configuration

Getting an API Key

Key File

External APIs (Optional)

Thingiverse

Cults3D

Research Tools

Integration Testing

Test All Integrations

Verify in Container

Related Documentation