Skip to content
The Daily Repo
Menu
Search repos

Prescott-Data / jarviscore-framework

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.

Why this job: topics: agent-framework, multi-agent-systems, llm-agents, ai-agents, agents, agentic-ai

Open on GitHub (opens in new tab)Website (opens in new tab)
Two blueprints side by side. A, a central orchestrator: one box that plans, routes and holds state, with arrows out to scout, coder, browser and writer, marked as a single point of failure; if it stops, everything stops. B, the JarvisCore peer mesh: scout, coder, browser and writer connected to each other, an analyst shown stopped, each live peer linked to a shared ledger of steps, claims and results in Redis, and coder and browser linked to the Nexus zero-trust broker; a node stops and a peer claims its step from the ledger.
Watch the video (opens in new tab)
Video from the Prescott-Data/jarviscore-framework READMEWatch the video (opens in new tab)
Watch the JarvisCore overview on YouTube (2:07): an open-source runtime for AI agents that work as a mesh of equal peers.
The model at the top, connected to four groups of components. Write: AutoAgent (role, capabilities, prompt) and CustomAgent (plain Python in execute_task). Run: the Kernel (roles, budgets, done check) and the mesh (peers, shared ledger, pools). Remember and reach: memory (notes, episodes, Athena), Nexus (credentials placed at the call) and integrations (typed atoms, auth injected). People: people in the loop (approve once, then it runs) and rooms (shared files, Slack), new in 2.3. You write the first column; the runtime supplies the rest.
Play the recording: the JarvisCore Desk running a three-agent research mesh. The request is typed, the Scout searches the web and reads official docs, the Analyst compares the options and hands its comparison to the Writer, and the Writer delivers a decision brief with source links as a document.
1 of 6

What it does

Most agent frameworks put one orchestrator in charge: it plans, routes every request and holds the state, so the system is only as available as that one process. On a JarvisCore mesh, every agent is a peer. Peers find each other by what they can do (SWIM gossip), take steps from a shared ledger (Redis) and record what they did there, so when one stops, another claims its step and the work goes on. None of them holds a credential: the Nexus broker places it at the call.

Get started

No single install command. Read the steps in the README (opens in new tab)

Maintenance
Active
Last commit 13 days ago
Stars
124
On GitHub
Licence
Apache-2.0
Allows business use
Downloads for
None listed
No release files
Latest version
v3.0.1
Released 13 days ago
Built with
Python
98% of the code

Is it looked after?

Contributors
2
Last 90 days
Issues
47% closed
10 still open
Pull requests
29 merged
10 waiting
Releases
9
28 in the last year

Similar repos

Repos similar to this one
RepoStarsLast commitLicenceDownloads for
google / adk-python22,154todayApache-2.0No downloads
opencmit / alphora3484 months agoApache-2.0No downloads
embabel / embabel-agent4,550yesterdayApache-2.0No downloads
More: languages, topics and project record

Languages

  • Python 98%
  • CSS 1%
  • HTML <1%
  • JavaScript <1%

Topics

  • agent-framework
  • agentic-ai
  • agents
  • ai
  • ai-agents
  • aiagentframework
  • distributed-systems
  • llm-agents
  • llms
  • multi-agent-systems
  • python
  • workflow-engine

Project record

  • JobAgent frameworks
  • StatusListed
  • Commits535
  • Forks17
  • Watching0
  • Stars in 28 days+12
  • CreatedJan 7, 2026
  • First seen by usNov 7, 2026
  • CheckedNov 7, 2026

Type to search.

Try