Mantequilla-Soft – Weekly Progress Report #4

Welcome to another Mantequilla-Soft Weekly Report!

Oh boy, oh boy. We said we were going to slow down because we need to find more income sources, but this week the Mantequilla Soft team doubled down and pushed harder, aka, added more butter to the apps and tools to make them even smoother.

Special mention to @tibfox, who absolutely crushed it with the 3Speak Shorts this week. The whole team is super grateful and super excited about this new feature. More details below. He also worked on a new video player to fix some issues, and the whole team is very excited about what he is doing on the shorts, playlist, video editor, and other awesome features in 3Speak.

Having said that, we would also like to mention the efforts and contributions of @meno. This week, he had a bunch of sleepless nights and troubleshooted like never before. He decided to fix the encoder's code and the encoding gateway, and while there are still things to improve, there are already significant improvements. More details below, but long story short, encoding job assignments are going to be more fair, and the user experience will improve. Another thing he tackled was the IPFS servers that host the 3Speak videos. Together with @tibfox, he added a second Hot Node. Isn't that cool? Or should we say hot?

Another valuable member of Mantequilla Soft is @kesolink. This week, he has been very busy working on the audio side of things. Soon, we will be able to show more, but the idea is to have a section (and a UI) on the site where people can upload audio. These audios could be music, podcasts, or whatever, all using the embed technology that you can find on audio.3speak.tv.

Last but not least, @eddiespino. Yes, me. The person in charge of writing these reports and some other management stuff. Well, this week I was focused on Search Engine Optimization. This is in preparation for all the people we want to onboard to 3Speak (and to Hive). At the moment, the SEO of 3Speak is not great. We found some critical bugs and have already fixed some of them. Also, this week, @eddiespino (I'm not sure why I'm talking in third person here) helped @tibfox troubleshoot some bugs in the 3Speak shorts.

You can check the 3Speak Preview at:

https://preview.3speak.tv/

(This is like a beta version of 3Speak)

Summary of updates by contributors:

Below you can find a technical summary or report of what each contributor did this week.

This includes a short description of each task or milestone and its impact.

@eddiespino

SEO Improvements for 3Speak.tv

A full SEO foundation overhaul for the 3Speak frontend is in progress. This will help make the platform visible to search engines and improve link previews across social platforms.

This work will introduce core SEO infrastructure, including robots.txt, sitemap indexes, and dynamic sitemaps for videos and channels powered by the Pancreas API. Static and dynamic metadata support was added using a reusable SEO component, enabling per-page titles, descriptions, Open Graph tags, Twitter cards, and canonical URLs.

Structured data was also implemented using JSON-LD:

VideoObject schema on watch pages for rich video search results
Organization and Website schema on the homepage

Additionally, serverless sitemap generators were added via Vercel functions, allowing continuously updated video and channel sitemaps without manual maintenance.

Impact:

With this, 3Speak will be properly discoverable by search engines. This will support rich previews when sharing videos and will have a scalable SEO foundation for future indexing and growth.

Other contributions by @eddiespino this week:

Tested the 3Speak shorts on iPhone
Found a critical error on an encoded video and reported the issue here

@tibfox

Snapie Video Player:

This week, @tibfox continued improving the Snapie (3Speak) player with several enhancements focused on embedding flexibility and playback reliability.

New events and parameters were added to the player API, including orientation lock controls and iframe URL parameters such as mute and loop for richer embedding options. Picture-in-Picture support was also expanded with a new toggle command.

Internally, mute handling was made more reliable by tracking the intended mute state independently from the Video.js player state. This prevents unintended unmuting during source initialization and ensures mute behavior is correctly enforced before playback. Additional state reporting and debugging improvements were also added for better integration and troubleshooting.

Impact:

These updates make the Snapie player more robust and easier to integrate, improving embedded playback control while reducing edge-case issues around mute state and player behavior.

New 3Speak Player SDK - GitHub Repo

NPM Package: `@mantequilla-soft/3speak-player`.

This week, @tibfox also created the official Mantequilla-Soft npm presence and published the first public package: npmjs.com/package/@mantequilla-soft/3speak-player.

The package is a framework-agnostic HLS video player SDK for 3Speak that removes the need for iframes and enables native <video> playback using hls.js on desktop browsers and native HLS on Safari/iOS. It supports vanilla JavaScript as well as React, Vue, and Svelte integrations.

Key features include lightweight playback, automatic CDN fallback handling, iOS-optimized playback strategies, TypeScript support, and a rich player API with events, controls, and multi-player pooling for feeds and shorts-style experiences.

Impact:

This introduces a reusable developer SDK for 3Speak video playback, enabling external apps and integrations to embed 3Speak content more reliably while avoiding iframe limitations, especially on iOS.

Additional 3Speak Improvements by @tibfox

This week included a wide range of product improvements across reactions, shorts UX, playback controls, and overall platform polish.

A major addition was the introduction of video reactions and reaction chains, allowing users to upload or record reaction videos directly as Hive comments. The system builds a visual breadcrumb chain from the original video through intermediate reactions, with overlays in Shorts and a dedicated ReactionPlayer on the watch page for browsing and interacting with reaction threads.

The Shorts experience received significant upgrades, including gesture controls (tap, double-tap, long press), keyboard shortcuts, improved mute handling, smarter preloading and iframe cleanup, comment rendering improvements, and better navigation stability. A user-specific shorts feed and enhanced pagination logic were also introduced.

Voting UX was redesigned into a more responsive modal system with improved performance and smoother interaction. At the same time, a mobile-first FAB speed-dial added quick actions like vote, tip, playlist, and share.

Additional improvements included:

User-specific Shorts feeds and better Shorts history tracking
Reaction and reshare systems, including counts and UI indicators
Playback enhancements like Picture-in-Picture support and player resize modes
Watch page resilience improvements with Hive fallbacks and better error handling
Low-reputation comment handling and reputation utilities
UI consistency improvements across badges, thumbnails, overlays, and navigation
Performance optimizations in API access and caching
Authentication and provider improvements, including HiveSigner support via AIOHA
General polish across profiles, playlists, history, and mobile navigation

Impact:

These changes significantly expand interactivity on 3Speak, introducing social video reactions, improving Shorts usability, strengthening playback controls, and delivering broad UX and performance improvements across both mobile and desktop experiences.

Please test the 3peak Shorts here: https://preview.3speak.tv/shorts.

Send us as much feedback as you can.

More contributions by @tibfox this week:

3Speak Playlists
Follow feed on 3Speak
Video editor inside of 3Speak
A new HotNode called "hinode"
Replaced the temporary YouTube Shorts logo with a personalized 3Speak Shorts logo

@meno

Infrastructure and Gateway Improvements by @meno

This week focused heavily on reliability and scalability across the encoding pipeline, IPFS integration, and job distribution systems.

A major change was the introduction of a round-robin auto-assignment system in the encoder gateway. Jobs are now pre-assigned to active workers based on current load, eliminating race conditions where multiple encoders compete for the same job. This also improves fairness and reduces wasted retries, while faster reassignment logic shortens recovery time for stalled jobs.

Several gateway reliability fixes were also implemented, including:

Preventing duplicate job creation using atomic database constraints
Switching IPFS size checks to the local gateway to avoid external timeouts
Improved heartbeat tracking so idle encoders remain visible to the scheduler
Encoder timing adjustments to align with the new gateway behavior

On the infrastructure side, groundwork began for decoupling hotnodes from the supernode, including CID resolution checks and improved pin handling. This work supports moving thumbnail handling away from the supernode and toward more reliable external hosting (Hive image server), reducing thumbnail failure risk.

Impact:

These changes significantly improve encoding reliability, eliminate job distribution race conditions, and strengthen the overall stability of the video processing pipeline while laying the foundation for more resilient storage.

Additional contributions by @meno:

Updates on the Pancreas API to fix the fetching of data on the 3speak frontend
Helped @tibfox with his new Hot Node

@kesolink

3Speak Audio Platform – Frontend Progress

This week, @kesolink made strong progress on the upcoming audio experience for 3Speak, covering both upload and discovery.

On the Audio Studio side, the foundation for a full upload flow was implemented, including drag-and-drop support for common audio formats, improved macOS compatibility, upload progress tracking, and a built-in waveform player with seek support. The interface is designed to auto-detect audio metadata and support a complete publishing flow with title, Markdown description, and shareable output details.

The Audio Feed also moved closer to a real browsing experience, transitioning away from placeholder data toward live API integration. It is being built to support pagination, loading, and error states, and a responsive grid layout, where each track will have its own dedicated player page.

Impact:

This work lays the groundwork for audio on 3Speak, establishing the core frontend architecture for uploads and discovery. Future steps include blockchain login, engagement features, and deeper audio interactivity before full release.

Additional contributions by @kesolink: