378 karma · joined June 10, 2025
The documentation, built from repo, is available here: https://deepseek-harness.github.io/deepseek-harness/en/guide... (I find the development and reference sections easier to read and navigate)
Any feedback on how to make it less "shitty"? I feel like doing some vibe coding tonight.
The article makes strong points, includes real data and quotes, shows proof of work (sampling 100 Q&A), so does that even matter at this point? This doesn't feel like "slop" to me at all.
Might be helpful:
"prettier": {
"allowed": true
},> Fun fact: Manus is currently SOTA on the Remote Labor Index (RLI) benchmark that @scale_AI and @ai_risks released earlier this year.
Source: https://x.com/alexandr_wang/status/2005766469771223106
If you've been following Manus and their work on context engineering, or have used the product, that line doesn't come off as satire IMO.
> [1] No, you can do whatever you want with it since it’s MIT
But if you use it in a commercial project, they'll likely shame you if you don't pay - like what happened to OpenAI with Agent Builder:
> [2] Hey @OpenAI :) We just saw that you are using our open source library React Flow We offer startup discount codes :) Let us know if you are interested
Personally, I’d prefer them to use a dual license, but I understand that it would likely create unnecessary hurdles for devs who just want to try out the library.
They explained their reasoning here: https://xyflow.com/blog/asking-for-money-for-open-source