seq-assist
Seqular's Digital Assistant
seq-assist is Knixus Technologies' Mastodon bot framework — the digital assistant living on network.seqular.net that answers questions, provides technical support, announces products, and represents your organization 24/7 on the Seqular network. Every Seqular Core Team member gets their own fully configured instance, free and unlimited.
Your organization's voice on the Seqular network — always on, always consistent
seq-assist is a Mastodon bot framework that gives every organization on Seqular a dedicated AI-powered digital assistant. It monitors mentions in real time, generates contextually aware responses using your chosen LLM, and handles everything from open-source coding questions to product support and company announcements.
NullPool explicitly when creating your async engine — it prevents connection sharing across event loop boundaries. Pass poolclass=NullPool to create_async_engine(). Each request gets a fresh connection — no loop leaks.
Meet Qwen — Knixus's digital PM assistant, powered by seq-assist
Knixus Technologies runs its own seq-assist instance as Qwen — the digital project manager and community assistant active on network.seqular.net. Qwen answers technical questions, helps developers navigate Knixus products, and represents the Knixus team on the Seqular network.
Everything you need to run a production Mastodon assistant
Monitors Mastodon notifications continuously. Every mention triggers an async processing pipeline — content parsing, user validation, duplicate detection, and AI response generation — in seconds.
Supports OpenRouter, MiniMax, and Alibaba/Qwen out of the box. Active model is configured per instance in the admin panel — switch providers without touching code. API keys stored securely in the database, not in environment files.
System prompts are Jinja2 templates with token budgeting. Inject dynamic context — user history, conversation thread, organization data. Define exactly how your bot thinks, what it knows, and how it responds.
Five thread-safe background services run concurrently: Welcome messages for new followers, Follow-thank replies, Timeline scanner, Follower tracker, and Translation proxy. Start, stop, and restart each independently from the admin panel.
Automatic detection of bot accounts prevents infinite reply loops. Visibility filtering ensures the bot only responds to public mentions. Duplicate prevention guarantees each mention is processed exactly once.
LibreTranslate-compatible API endpoint powered by the active LLM. Mastodon's built-in translation feature routes through seq-assist — no external translation service required. Results cached for performance.
How Seqular Core Team members deploy seq-assist
Each organization configures their instance for a different purpose. The same framework, infinite configurations.
Answer product questions, handle support requests, and guide users through troubleshooting — automatically, at any hour. Configure the bot with your product knowledge and support procedures.
New release? Feature update? Use the Timeline Scanner service to engage the community proactively. Configure welcome messages for new followers to introduce your product to every new community member automatically.
For open-source projects on Seqular — answer coding questions, explain APIs, point to documentation, and guide contributors. Configure with your project's codebase context and contribution guidelines.
Guide new team members through your technology stack, answer infrastructure questions, and provide consistent onboarding information — without requiring a senior engineer to be available.
As demonstrated by Knixus's own Qwen instance — configure seq-assist as a project management assistant that answers questions about ongoing work, tooling, and processes for the development team.
The built-in translation proxy lets your bot handle questions in multiple languages through Mastodon's native translation feature. Reach the entire Seqular community regardless of language barriers.
Full control — no configuration files, no restarts
Every setting is managed through the web admin panel. Change AI models, edit prompts, manage services, and review logs — all without touching the server.
Knixus Technologies provides seq-assist to every Seqular Core Team member at no cost and with no usage limits — provisioned through Knixus cloud as a dedicated instance. Your own Mastodon account, your own configuration, your own data. Zero shared infrastructure between organizations.
Built on solid, production-proven foundations
Async-first web framework powering both the API and the bot event loop. Database-free startup with lazy module loading.
First-class Mastodon API client. Real-time notification streaming, posting, follow management, and credential handling.
Async ORM with QueuePool connection management. All configuration — AI models, prompts, settings — stored in the database.
WSGI-to-ASGI bridged Flask panel mounted on the FastAPI app. Web-based control without a separate process.
Template-based prompt system with token budget enforcement, context injection, and fallback rendering.
Schema version management with auto-initialization. Database set up automatically on first run — no manual migration step needed.
Give your organization a voice
on the Seqular network
seq-assist is available free and unlimited to all Seqular Core Team members via Knixus cloud. Your instance, your configuration, your community presence.