Village nerd with some tools • Founder @prescottdata • Author https://xitter.cf/t.co/l4EoAquzRk • Creator of @TurnLang
London, England
Joined May 2022
- Tweets517
- Following101
- Followers65
- Likes473
Pinned Tweet
My hope for JarvisCore is that it shows a completely different way to build AI agents. Not as an experiment, but as an approach we've proven.
We open-sourced it a few months ago and kept building. Today it launches on Product Hunt. Here's what makes it different:
• No master agent. Agents are equal peers in a decentralised P2P mesh
• No MCP servers needed
• Zero-trust credentials
• Bounded memory (agents forget on a curve)
If you're building agents, I'd love to hear what's breaking for you.
this week JarvisCore goes multiplayer.
agents are already equal peers in a P2P mesh, so we added people as peers in the very mesh. we're also dropping a blueprint implementation so you can see how it's built and fork yours (if interested leave me a comment)
come build and tell me what breaks
github.com/Prescott-Data/jar…
If you found value from JarvisCore , a ⭐ helps other developers find it.
abandoning everything to be like bro, bro won
Chinese bride gifts her partner a flagship GPU at wedding — he breaks down in tears
She handed him an ASUS ROG Astral RTX 5090 that costs around $12,000
Muyukani Kizito retweeted
I am on a project that started out as an internal tool, that is now open source. I've had to rethink the architecture on the go. I'll be having a chat with some internet strangers on what changed, why and how at 11 am (UTC+3)
discord.gg/gvry5NShz?event=1…
Muyukani Kizito retweeted
No master agent. No MCP servers. No keys in .env.
Agents as equal peers in a P2P mesh with zero-trust credentials and memory that stays sharp without growing forever.
My hope for JarvisCore is that it shows a completely different way to build AI agents. Not as an experiment, but as an approach we've proven.
We open-sourced it a few months ago and kept building. Today it launches on Product Hunt. Here's what makes it different:
• No master agent. Agents are equal peers in a decentralised P2P mesh
• No MCP servers needed
• Zero-trust credentials
• Bounded memory (agents forget on a curve)
If you're building agents, I'd love to hear what's breaking for you.
Hello founders i'm sharing with you this open-source agent framework where agents are equal peers in a decentralised P2P mesh. But quite frankly the part i'm proud of the most because it's made tool wiring much easier is that you don't need any MCP wiring
github.com/Prescott-Data/jar…
Engage with us on PH producthunt.com/products/jar… via @producthunt
JarvisCore is open-source: github.com/Prescott-Data/jar…
Last week we shipped persistent source workspaces, this week we've shipped long running research for your agents.
Thursday is for office hours with developers
I’ll be walking you through what brought me to build nexus-cli in today’s office hour session
discord.gg/vrUmpy8Cd?event=1…
I am finding giving GPT 5.6 Sol deadlines is cutting down on my token spend e.g “no being performative to try rob me my beautiful tokens get this done in under 10 mins”
anyone else noticed new models have performative progress to burn more tokens and are always stretching routine work across needless turns?
I threatened GPT 5.6 Sol today after burning my tokens for hours and it gave me results in 10
Muyukani Kizito retweeted
JarvisCore 1.13 now gives agents durable source workspaces.
Point a Mesh at a GitHub repo and the agent works from an immutable snapshot. Each step gets its own isolated copy-on-write workspace. Every edit is hash-guarded, so a stale edit is rejected instead of overwriting newer code. Commands and file changes come back as verified receipts, and only the accepted diff is kept.
You can also connect your own sources, like archives, datasets or document collections, through the same adapter interface.
JarvisCore 1.13 now gives agents durable source workspaces! go try it out and leave us some feedback.
Catch what this latest version entails in our blog here developers.prescottdata.io/b…
