OmnicoreAgent

v1.0.0Coding Agentsstable

Open Python agent harness for production AI apps: tools, MCP, memory, workspace, telemetry, subagents, background tasks, and OmniServe APIs.

agentagent-harnessai-agentsbackground-tasksfastapi
Share:
238
Stars
0
Downloads
0
Weekly
0/5

What is OmnicoreAgent?

OmnicoreAgent is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to open python agent harness for production ai apps: tools, mcp, memory, workspace, telemetry, subagents, background tasks, and omniserve apis.

Open Python agent harness for production AI apps: tools, MCP, memory, workspace, telemetry, subagents, background tasks, and OmniServe APIs.

This server falls under the Coding Agents category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • Open Python agent harness for production AI apps: tools, MCP

Use Cases

Build production AI apps with tools
Manage agent memory and workspace
Handle background tasks
LicenseMIT
Languagepython
Versionv1.0.0
UpdatedMay 21, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

Manual Installation

npx omnicoreagent

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use OmnicoreAgent

OmniCoreAgent is an open Python agent harness built for production AI applications, providing a rich set of runtime features that go beyond basic LLM calls: parallel batch tool execution, multi-tier memory backends (in-memory, Redis, MongoDB, PostgreSQL), automatic context management with summarisation, dynamic subagents with isolated contexts, MCP native tool integration, and an optional OmniServe REST/SSE API layer. It is designed for engineers who need a reliable, observable, and extensible foundation for deploying AI agents in production without rolling their own tool dispatch, loop detection, or context truncation logic.

Prerequisites

  • Python 3.9 or later installed
  • An LLM provider API key set as LLM_API_KEY (OpenAI, Anthropic, or any OpenAI-compatible provider)
  • pip or uv for package installation
  • Optional: Redis, PostgreSQL, or MongoDB for persistent memory backends
  • An MCP-compatible client if using OmniCoreAgent as an MCP server
1

Install the core package

Install OmniCoreAgent from PyPI. The base install includes in-memory storage and local file workspace. Install optional extras for production backends such as Redis, PostgreSQL, or MongoDB memory, S3 workspace storage, or OTLP telemetry.

pip install omnicoreagent
# With all optional extras:
pip install "omnicoreagent[all]"
# Individual extras:
pip install "omnicoreagent[redis]"     # Redis memory backend
pip install "omnicoreagent[postgres]"  # PostgreSQL/SQL memory
pip install "omnicoreagent[serve]"     # OmniServe REST/SSE API
pip install "omnicoreagent[otel]"      # OTLP telemetry
2

Set your LLM API key

Export your LLM provider API key. OmniCoreAgent uses litellm-style provider/model identifiers so you can point it at OpenAI, Anthropic, Ollama, or any compatible endpoint.

export LLM_API_KEY=your_api_key_here
3

Create and run a basic agent

Instantiate an OmniCoreAgent with a name, system instruction, and model config. Call agent.run() with your task and a session_id for memory continuity across calls. Always call agent.cleanup() when done to release resources.

import asyncio
from omnicoreagent import OmniCoreAgent

agent = OmniCoreAgent(
    name="assistant",
    system_instruction="You are a helpful research assistant.",
    model_config={"provider": "openai", "model": "gpt-4o"},
)

async def main():
    result = await agent.run(
        "Research the top 3 open-source agent runtimes in 2025.",
        session_id="research-session",
    )
    print(result["response"])
    await agent.cleanup()

asyncio.run(main())
4

Configure advanced agent options

Pass an agent_config dictionary to tune runtime behaviour: maximum steps, tool call timeout, guardrail mode, context management strategy, and workspace file offloading.

agent_config = {
    "max_steps": 15,
    "tool_call_timeout": 30,
    "enable_workspace_files": True,
    "guardrail_mode": "full",
    "context_management": {"enabled": True},
    "tool_offload": {"enabled": True},
    "enable_advanced_tool_use": True,
}
5

Add MCP server tools to the agent

OmniCoreAgent supports native MCP tool connections. Pass a list of MCP server commands or configs so the agent can call external MCP tools during execution alongside its built-in local tools.

6

Configure as an MCP server for Claude Desktop

Add OmniCoreAgent to your Claude Desktop configuration to use it as an MCP server. The server is installed via uvx for zero-friction deployment.

{
  "mcpServers": {
    "omnicoreagent": {
      "command": "uvx",
      "args": ["omnicoreagent"],
      "env": {
        "LLM_API_KEY": "your_api_key_here"
      }
    }
  }
}

OmnicoreAgent Examples

Client configuration

Claude Desktop configuration for OmniCoreAgent with LLM API key.

{
  "mcpServers": {
    "omnicoreagent": {
      "command": "uvx",
      "args": ["omnicoreagent"],
      "env": {
        "LLM_API_KEY": "your_api_key_here"
      }
    }
  }
}

Prompts to try

Example prompts for working with OmniCoreAgent's production-grade agent harness.

- "Research the latest developments in LLM agent frameworks and return a structured comparison"
- "Run a multi-step analysis: fetch data, clean it, and produce a summary report with key metrics"
- "Spawn a subagent to handle the database queries while the main agent processes the results"
- "Use the Redis memory backend to recall what we discussed in our last session about project requirements"
- "Execute these 10 tool calls in parallel and aggregate the results"

Troubleshooting OmnicoreAgent

ModuleNotFoundError for redis, psycopg2, or motor when using memory backends

Install the appropriate optional extra: 'pip install omnicoreagent[redis]' for Redis, 'pip install omnicoreagent[postgres]' for PostgreSQL, or 'pip install omnicoreagent[mongodb]' for MongoDB. The base install only includes in-memory storage.

Agent hits max_steps limit without completing the task

Increase max_steps in your agent_config dictionary. For complex research tasks, 20–30 steps may be needed. Also enable context_management so the agent does not waste steps on context overflow handling.

Loop detection triggers prematurely, stopping the agent

OmniCoreAgent uses SHA256-backed signature detection to identify repeated tool call patterns. If your task legitimately requires calling the same tool with the same arguments multiple times, adjust the loop detection sensitivity in the agent_config or restructure your task to vary the tool inputs.

Frequently Asked Questions about OmnicoreAgent

What is OmnicoreAgent?

OmnicoreAgent is a Model Context Protocol (MCP) server that open python agent harness for production ai apps: tools, mcp, memory, workspace, telemetry, subagents, background tasks, and omniserve apis. It connects AI assistants to external tools and data sources through a standardized interface.

How do I install OmnicoreAgent?

Follow the installation instructions on the OmnicoreAgent GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.

Which AI clients work with OmnicoreAgent?

OmnicoreAgent works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is OmnicoreAgent free to use?

Yes, OmnicoreAgent is open source and available under the MIT license. You can use it freely in both personal and commercial projects.

Browse More Coding Agents MCP Servers

Explore all coding agents servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "omnicoreagent": { "command": "npx", "args": ["-y", "omnicoreagent"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use OmnicoreAgent?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides