Build agents as peers in a mesh network with zero-trust
JarvisCore is a runtime where AI agents operate as a fleet of equal peers. Agents discover one another by capability, execute tasks from a shared ledger, and authenticate to every external service through a zero-trust broker, never with their own keys.
Hi Product Hunt, I'm Muyukani, founder and CEO of Prescott Data. We build the infrastructure that makes AI usable in regulated industries, and JarvisCore is the open-source runtime underneath it.
What inspired it:
Every agent framework we tried had the same shape: a master agent orchestrating workers, MCP servers bolted on for tools, API keys in a .env file, and memory that either grows forever or gets truncated. That shape works in a demo. It breaks the moment agents run for real.
The problem we wanted to solve:
1. The master orchestrator is a single point of failure and a scaling ceiling. When it dies, the fleet dies.
2. Every MCP server brings its own provider auth wiring. Ten providers means ten auth headaches.
3. Credentials sitting in .env files end up in agent context, prompts and traces.
4. Memory is either unbounded, so cost and noise grow forever, or lossy, so agents reason from fragments.
How it evolved:
We ran our own agents on JarvisCore with real budgets and fixed what broke, one release at a time. That led to four decisions:
1. No master agent. Agents are equal peers in a decentralised P2P mesh. They discover each other, claim work, and pick up a failed peer's work. No single point of failure, and you scale by adding peers.
2. No MCP servers. Agents write their own tools, and JarvisCore orchestrates provider authentication natively, so an agent can authenticate across any provider.
3. Zero-trust credentials. Our broker, Nexus, injects credentials at call time. Agents never see a key, and nothing depends on .env files.
4. Bounded, lossless memory. Agents forget on a curve: what gets used stays, what doesn't fades. Memory stays bounded without truncating what the agent needs.
Why you should use it:
If you're building agents that have to run unattended, across many systems, with real credentials, JarvisCore gives you all four in one Apache-2.0 package. Start with one AutoAgent in three attributes, then scale the same code across machines.
pip install jarviscore-framework
If you run agents beyond demos, I'd love to hear what breaks first for you.
About JarvisCore on Product Hunt
“Build agents as peers in a mesh network with zero-trust”
JarvisCore launched on Product Hunt on October 2nd, 2026 and earned 101 upvotes and 9 comments, placing #17 on the daily leaderboard. JarvisCore is a runtime where AI agents operate as a fleet of equal peers. Agents discover one another by capability, execute tasks from a shared ledger, and authenticate to every external service through a zero-trust broker, never with their own keys.
On the analytics side, JarvisCore competes within Open Source, Artificial Intelligence, GitHub and Tech — topics that collectively have 1.2M followers on Product Hunt. The dashboard above tracks how JarvisCore performed against the three products that launched closest to it on the same day.
Who hunted JarvisCore?
JarvisCore was hunted by Muyukani Kizito. A “hunter” on Product Hunt is the community member who submits a product to the platform — uploading the images, the link, and tagging the makers behind it. Hunters typically write the first comment explaining why a product is worth attention, and their followers are notified the moment they post. Around 79% of featured launches on Product Hunt are self-hunted by their makers, but a well-known hunter still acts as a signal of quality to the rest of the community. See the full all-time top hunters leaderboard to discover who is shaping the Product Hunt ecosystem.
For a complete overview of JarvisCore including community comment highlights and product details, visit the product overview.
Muyukani Kizito