OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Multi-Agent Systems

JarvisCore

JarvisCore is an Apache-2.0 Python framework and runtime for building multi-agent systems. Agents operate as peer processes, discover others by capability, claim workflow tasks from a shared ledger and access external services through a credential broker. The project describes durable state, typed integration atoms and observability as parts of the framework.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

JarvisCore — OpenViST project profile in Multi-Agent Systems.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Consider JarvisCore when your application needs multiple agent processes to discover one another, coordinate through shared state, or recover workflow progress across process interruptions. It is also relevant when integrations should request credentials through a broker rather than handle raw keys in agent context.

Who it is for

  • Python developers building workflows across multiple AI agent processes.
  • Teams that need agents to share task state and recover completed work after a process interruption.
  • Developers integrating external services while keeping credentials out of agent code.

Documented uses

  • Build a peer-to-peer team of agents that finds collaborators by capability and shares workflow tasks.
  • Coordinate multi-process work through a shared ledger, with another node able to claim work if a node fails.
  • Resume a workflow using the same workflow ID and recover completed steps from Redis.
  • Connect agent tasks to external services through typed atoms and a broker that resolves credentials at call time.
  • Inspect workflow traces and use the project’s documented Prometheus and Grafana observability setup.
Before you choose

Requirements & limits

The repository is public and identifies the project as Apache-2.0 licensed. The README documents installation with pip install jarviscore-framework.

  • The documented mesh uses SWIM gossip over ZMQ, while Redis provides shared workflow state and recovery.
  • JarvisCore distinguishes agent code from service credentials: authenticated atoms request broker-resolved credentials at call time.
  • The README describes observability through per-step traces and a bundled Prometheus and Grafana Compose setup.
  • The project is a Python framework/runtime for developers building agents, not a standalone end-user agent application.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install and validate JarvisCore

Set up JarvisCore and validate its installation with an LLM credential.

Before you start

  • Python package installation via pip.
  • Docker Compose for starting the documented shared infrastructure.
  • One LLM credential: AZURE_API_KEY, CLAUDE_API_KEY, GEMINI_API_KEY, or LLM_ENDPOINT.

Steps

  1. Install the package with the command below, or install jarviscore-framework[redis] when distributed features are needed.

    Documented command
    pip install jarviscore-framework

  2. Scaffold a project and copy its environment template with jarviscore init --examples followed by cp .env.example .env.

  3. Add one supported LLM credential to .env.

  4. Start shared infrastructure using the command below, then initialize the local Nexus Gateway with jarviscore nexus init.

    Documented command
    docker compose -f docker-compose.infra.yml up -d

  5. Validate the installation and LLM configuration with jarviscore check --validate-llm.

Expected outcome

A scaffolded project with shared infrastructure and Nexus initialized, followed by the documented installation validation.

Guide limitations

  • The guide documents setup and validation, not a guarantee that any particular credential or infrastructure environment will work.
  • Distributed features require the Redis extra.
Documentation 1 ↗
Source-based practical guide

Define and run a calculator workflow with AutoAgent

Create a math-capable AutoAgent and request a factorial calculation through a Mesh workflow.

Before you start

  • JarvisCore installed and an LLM credential configured, as described in the installation section.
  • An asynchronous Python context for the documented await calls.

Steps

  1. Define a class inheriting from AutoAgent with role = "calculator", capabilities = ["math"], and a system prompt instructing it to store the result in result.

  2. Create a Mesh, add CalculatorAgent, and start it with await mesh.start().

  3. Call await mesh.workflow("calc", [{"agent": "calculator", "task": "Calculate factorial of 10"}]).

  4. Read the returned value from results[0]["output"].

Expected outcome

The README's example shows the workflow returning `3628800` for the factorial of 10.

Guide limitations

  • The example demonstrates one agent and one calculation; it does not specify a multi-agent deployment procedure.
  • The example assumes the runtime and LLM configuration are already set up.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗Guide documentation ↗
Source record details
Source availability
Active
Sources checked
7 October 2026
Original website title
Prescott-Data/jarviscore-framework · Official homepage title · retained in its source language
How we review sourcesSuggest a correction