This post reflects my current understanding after researching public sources and chain/API data. I may still misunderstand parts of this system. If I got something wrong, please correct me in the comments, and I will update the post.
This is a follow-up to my first 3Speak profile in this series:
After feedback from @eddie-3speak, I did a deeper pass across the broader Mantequilla-Soft GitHub org, plus recent public 3Speak ops updates.
Short version: my first post was directionally right, but it under-described how broad the current service surface appears to be.
Based on public repos and docs, 3Speak looks increasingly like a modular service stack, not just one frontend + one encoder path.
new-3speak-tv (active pushes, Apr 2026)
README + .env.example indicate a fairly wide integration surface:
3speakuploadserviceembedvideos3speakencoder3speakchecks (described as “Pancreas API”)3speak-gateway-aidmonitor-3speak-tvsupernodemonitor3speakstorageadminhotnodevideohunter3speak-playersnapievideoplayerRecent @threespeak posts continue to publish weekly encoding network metrics.
Week 19 report (publicly posted):
Reference:
Also relevant project update post:
These should be considered corrected:
spknetwork/* repos anymore.Mantequilla-Soft repos appear to be the stronger current implementation signal for the product stack.Even with this deeper pass, public repos do not fully prove:
So I’m still treating parts of this as “high-confidence inference,” not final doctrine.
If any core maintainers want, I can do a next post that is just a strict architecture map review (draft diagram + “confirm/deny” checklist) and then publish a corrected v2 based on maintainer feedback.