frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Show HN: MUP – Interactive UI inside LLM chat, so anyone can use agentic AI

https://github.com/Ricky610329/mup
1•Ricky_Tsou•1h ago
Agentic AI is powerful, but most people never experience it — it's trapped behind text commands and dev tools.

MUP (Model UI Protocol) lets you embed interactive UI directly in LLM chat. Each MUP is just a single .html file. The same functions can be triggered by the user (clicking a button) or by the LLM (function call). Both sides see each other's actions in real time.

The repo includes a PoC host and 9 example MUPs. Demo mode lets you interact with the UI side without an API key. Add an OpenAI key to see full LLM-UI collaboration.

Demo videos in the README show things like: drawing pixel art then charting its colors, a camera that captures a scene and the LLM recreates it, making beats on a drum machine with the LLM.

I'd love feedback on the protocol design.

Comments

newexpand•1h ago
Interesting approach. Making agentic AI accessible through visual UI instead of text commands is a real gap right now.

The dual interaction model — where both the user and the LLM can trigger the same functions — is a nice design choice. It avoids the "watch the AI work" problem where you're just a spectator.

Curious about the protocol design: how do you handle conflicts when the user and LLM try to act on the same element simultaneously? And is there a way for MUPs to communicate with each other, or is each one isolated?

Ricky_Tsou•1h ago
Thanks!

MUP is just a protocol — it defines the format and communication between a MUP and its host, but it doesn't dictate how the host handles these things.

Conflict resolution depends on the host implementation. Different applications may need different strategies (queuing, locking, last-write-wins, etc.) — that's a host-level decision, not a protocol-level one.

Same for MUP-to-MUP communication. The protocol keeps each MUP isolated by design, but a host could absolutely build a coordination layer on top. In our PoC, the LLM acts as the orchestrator between MUPs, but that's just one approach.

The spec intentionally stays out of these decisions so it can work across different hosts and use cases.

Automating Forecasting Question Generation and Resolution for AI Evaluation

https://arxiv.org/abs/2601.22444
1•nbosse•1m ago•0 comments

The Los Angeles Aqueduct Is Wild [Practical Engineering]

https://www.youtube.com/watch?v=PhW2BjFQCZM
1•Jgrubb•2m ago•0 comments

Blocking the Internet Archive Won't Stop AI, but It Will Erase the Web's History

https://www.eff.org/deeplinks/2026/03/blocking-internet-archive-wont-stop-ai-it-will-erase-webs-h...
1•doener•3m ago•0 comments

3 months in production: Architecture of an autonomous AI pipeline

https://gammavibe.com/updates/autonomous-startup-generator-architecture/
1•digitalhobbit•4m ago•1 comments

Coding with agents feels like a chess simul

https://tobeva.com/articles/chess-simul/
1•pbw•4m ago•0 comments

Show HN: Hubcap – A single Go binary to automate Chrome via CDP

https://tomyandell.dev/blog/introducing-hubcap
3•tomyandell•5m ago•0 comments

When Humanoid Robots Come to a Small Town Factory in South Carolina

https://www.wsj.com/business/south-carolina-schaeffler-plant-robots-d56c91d0
1•JeanKage•5m ago•0 comments

Lewis 1.0 – 8B model trained on AI social data beats Sonnet on personality dvgnc

https://github.com/swarmgram/swarmgrampublic
2•swarmgram•6m ago•0 comments

Drawvg Filter for FFmpeg

https://ayosec.github.io/ffmpeg-drawvg/
1•nolta•6m ago•0 comments

Cuba's power system suffers total collapse

https://www.cnn.com/2026/03/16/americas/cuba-power-grid-collapse-intl-latam
1•randycupertino•6m ago•1 comments

Microsoft's 'unhackable' Xbox One has been hacked by 'Bliss'

https://www.tomshardware.com/video-games/console-gaming/microsofts-unhackable-xbox-one-has-been-h...
2•crtasm•6m ago•0 comments

The serious part-time unofficial hobby of videogame fan-translation

https://medium.com/@hilltopmailer/the-deadly-serious-part-time-unofficial-hobby-of-videogame-fan-...
1•AdmiralAsshat•7m ago•0 comments

What 100M Volts Do to the Body and Mind

https://www.theatlantic.com/magazine/2026/04/lightning-strike-survivors-body-mind/686057/
2•gmays•9m ago•0 comments

Large craters offer clues to the origin of asteroid 16 Psyche

https://phys.org/news/2026-03-large-craters-clues-asteroid-psyche.html
2•Brajeshwar•9m ago•0 comments

Praxis – an AI-native intermediate language for agentic workflows

https://github.com/cssmith615/praxis
2•cssmith615•10m ago•0 comments

Sparkling water mitigates cognitive fatigue during prolonged esports play

https://www.sciencedirect.com/science/article/pii/S2451958826000175
1•PaulHoule•11m ago•0 comments

Filtering through all the spam/crap on Amazon

https://bindiving.com/
1•cazzer•11m ago•0 comments

OpenReview: Open-source, self-hosted AI code review bot powered by Vercel

https://github.com/vercel-labs/openreview
4•nateb2022•12m ago•1 comments

Show HN: MCP server that turns local files into a shareable link instantly

https://github.com/file-kiwi/filekiwi-mcp-server
1•hwovh•12m ago•0 comments

Worse than kids stomping that pig's bladder (2023)

https://taylor.town/pigs-bladder
1•surprisetalk•15m ago•0 comments

Features of Fully-Fledged card systems (2017)

https://thewrongtools.wordpress.com/2017/10/15/features-of-fully-fledged-card-systems/
1•surprisetalk•15m ago•0 comments

Stop Complying

https://gomakethings.com/stop-complying/
2•surprisetalk•15m ago•0 comments

Reasoning about performance (in the context of search) [video]

https://www.youtube.com/watch?v=80LKF2qph6I
1•surprisetalk•15m ago•0 comments

I built my own AI powered online learning tool

https://github.com/symbiont-ai/docent
1•chiefsymbiont•15m ago•0 comments

800 Lines of Python found the Sun's rotation from free NASA data

https://github.com/SaulVanCode/protoscience-nasa-experiments
2•strujillo•16m ago•0 comments

LLMs – How did they get so good?

https://www.manhattanmetric.com/blog/2026/03/how-did-llms-get-so-good
1•jballanc•16m ago•0 comments

The Former Academic Guiding OpenAI's Trillion-Dollar AI Buildout

https://www.bloomberg.com/news/articles/2026-03-11/the-former-academic-guiding-openai-s-trillion-...
2•gmays•17m ago•0 comments

Show HN: Need is a CLI tool discovery as an MCP server

https://www.agentneeds.dev/
3•schreibertuc•17m ago•0 comments

Third GE Vernova blade break at wind farm powering Google

https://www.windpowermonthly.com/article/1951775/third-ge-vernova-blade-break-wind-farm-powering-...
1•genve•18m ago•0 comments

Firefighting Drones Head to Aspen

https://arstechnica.com/gadgets/2026/03/firefighting-drones-head-to-aspen-can-they-suppress-a-bla...
1•LorenDB•19m ago•0 comments