JarvisCore

Orchestrate multi-agent systems with peer-to-peer coordination, unified memory, and built-in auth.

Features

• Agent Profiles — AutoAgent auto-generates and executes function tools under Kernel supervision; CustomAgent is bring-your-own-logic for advanced use cases
• Human In the Loop (HITL) — Pause and resume agent execution for human review and approval
• P2P Mesh — Agent discovery, peer-to-peer communication, and cognitive discovery via SWIM protocol + ZMQ
• Workflow Orchestration — Dependencies, context passing, multi-step pipelines with crash recovery
• Distributed Autonomous Workers — Mesh claims workflow steps without hardcoding step IDs
• UnifiedMemory — EpisodicLedger, LongTermMemory, RedisMemoryAccessor, WorkingScratchpad
• Context Distillation — TruthContext, TruthFact, Evidence models for shared agent knowledge
• Nexus OSS Auth — Full OAuth flow via requires_auth=True; no boilerplate in agents
• Telemetry / Tracing — TraceManager (Redis + JSONL), Prometheus step metrics
• Integrations — FastAPI, aiohttp, and other async frameworks via jarviscore.integrations

Installation

pip install jarviscore-framework

# With Redis support (required for distributed features)
pip install "jarviscore-framework[redis]"

# With Prometheus metrics
pip install "jarviscore-framework[prometheus]"

# Everything
pip install "jarviscore-framework[redis,prometheus]"

Setup

# Initialize project
python -m jarviscore.cli.scaffold --examples
cp .env.example .env
# Add your LLM API key to .env

# Start Redis (required for mailbox, memory, distributed workflows)
docker compose -f docker-compose.infra.yml up -d

# Validate
python -m jarviscore.cli.check --validate-llm
python -m jarviscore.cli.smoketest

Quick Start

AutoAgent

from jarviscore import Mesh
from jarviscore.profiles import AutoAgent

class CalculatorAgent(AutoAgent):
    role = "calculator"
    capabilities = ["math"]
    system_prompt = "You are a math expert. Store result in 'result'."

mesh = Mesh(mode="autonomous")
mesh.add(CalculatorAgent)
await mesh.start()

results = await mesh.workflow("calc", [
    {"agent": "calculator", "task": "Calculate factorial of 10"}
])
print(results[0]["output"])  # 3628800

CustomAgent (Workflow Mode)

from jarviscore import Mesh
from jarviscore.profiles import CustomAgent

class ProcessorAgent(CustomAgent):
    role = "processor"
    capabilities = ["processing"]

    async def execute_task(self, task):
        data = task.get("params", {}).get("data", [])
        return {"status": "success", "output": [x * 2 for x in data]}

mesh = Mesh(mode="distributed", config={"bind_port": 7950, "redis_url": "redis://localhost:6379/0"})
mesh.add(ProcessorAgent)
await mesh.start()

results = await mesh.workflow("demo", [
    {"agent": "processor", "task": "Process", "params": {"data": [1, 2, 3]}}
])
print(results[0]["output"])  # [2, 4, 6]

Infrastructure Stack

Every agent receives the full infrastructure stack automatically — no wiring required.

Feature	Injected as	Enabled by
Blob storage	self._blob_storage	STORAGE_BACKEND=local (default)
Context distillation	TruthContext, ContextManager	automatic
Telemetry / tracing	TraceManager	automatic (PROMETHEUS_ENABLED for metrics)
Mailbox messaging	self.mailbox	REDIS_URL
Function registry	self.code_registry	automatic (AutoAgent)
Kernel OODA loop	Kernel (AutoAgent internals)	automatic (AutoAgent)
Distributed workflow	WorkflowEngine	REDIS_URL
Nexus OSS auth	self._auth_manager	requires_auth=True + NEXUS_GATEWAY_URL
Unified memory	UnifiedMemory, EpisodicLedger, LTM	REDIS_URL
Auto-injection	all of the above	automatic

class MyAgent(CustomAgent):
    requires_auth = True   # → self._auth_manager injected

    async def setup(self):
        await super().setup()
        # All infrastructure already injected — no __init__ wiring needed
        self.memory = UnifiedMemory(
            workflow_id="my-workflow", step_id="step-1",
            agent_id=self.role,
            redis_store=self._redis_store,
            blob_storage=self._blob_storage,
        )

    async def execute_task(self, task):
        # Save artifact
        await self._blob_storage.save("results/output.json", json.dumps(result))

        # Notify another agent
        self.mailbox.send(other_agent_id, {"event": "done", "workflow": "my-workflow"})

        # Log to episodic ledger
        await self.memory.episodic.append({"event": "step_complete", "ts": time.time()})

        return {"status": "success", "output": result}

Production Examples

All examples require Redis (docker compose -f docker-compose.infra.yml up -d).

Example	Mode	Profile
Ex1 — Financial Pipeline	autonomous	AutoAgent
Ex2 — Research Network (4 nodes)	distributed	AutoAgent
Ex3 — Support Swarm	p2p	CustomAgent
Ex4 — Content Pipeline	distributed	CustomAgent
Investment Committee	autonomous	AutoAgent + CustomAgent

# Ex1: Financial pipeline (single process)
python examples/ex1_financial_pipeline.py

# Ex2: 4-node distributed research network
python examples/ex2_synthesizer.py &       # Start seed first (port 7949)
python examples/ex2_research_node1.py &    # port 7946
python examples/ex2_research_node2.py &    # port 7947
python examples/ex2_research_node3.py &    # port 7948

# Ex3: Customer support swarm (P2P + optional Nexus OSS auth)
python examples/ex3_support_swarm.py

# Ex4: Content pipeline with LTM (sequential, single process)
python examples/ex4_content_pipeline.py

# Investment Committee: 7-agent workflow with web dashboard
cd examples/investment_committee
python committee.py --mode full --ticker NVDA --amount 1500000
# or: python dashboard.py  (web UI on http://localhost:8004)

Profiles

Profile	You Write	JarvisCore Handles
AutoAgent	System prompt (3 attributes)	Agent-generated function tools, Kernel OODA loop, sandboxed execution, repair, function registry
CustomAgent	on_peer_request() and/or execute_task()	Mesh, discovery, routing, lifecycle, full infrastructure stack

Execution Modes

Mode	Use Case
autonomous	Single machine, agent-generated function tools (AutoAgent)
p2p	Agent-to-agent communication, swarms (CustomAgent)
distributed	Multi-node workflows + P2P + Redis crash recovery

Integrations

JarvisCore is async-first. Best experience with async frameworks.

Framework	Integration
FastAPI	JarvisLifespan — 3-line setup via jarviscore.integrations.fastapi
aiohttp, Quart, Tornado	Manual lifecycle (see docs)
Flask, Django	Background thread pattern (see docs)

from fastapi import FastAPI
from jarviscore.profiles import CustomAgent
from jarviscore.integrations.fastapi import JarvisLifespan

class ProcessorAgent(CustomAgent):
    role = "processor"
    capabilities = ["processing"]

    async def on_peer_request(self, msg):
        return {"result": msg.data.get("task", "").upper()}

app = FastAPI(lifespan=JarvisLifespan(ProcessorAgent(), mode="p2p"))

Documentation

https://jarviscore.developers.prescottdata.io/

Guide	Description
Getting Started	5-minute quickstart
AutoAgent	Agent profiles, Kernel, distributed research network
CustomAgent	CustomAgent patterns, infrastructure stack, production walkthroughs
User Guide	Complete documentation including memory and auth
API Reference	Detailed API docs for all infrastructure classes
Configuration	Settings reference and environment variable guide
Troubleshooting	Common issues and diagnostics
Changelog	Full release history

Docs are also bundled with the package:

python -c "import jarviscore; print(jarviscore.__path__[0] + '/docs')"

Version

1.0.2

License

Apache 2.0 — see LICENSE for details.