Agenttier is an open-source agent orchestration harness and execution platform designed for hosting, managing, and scaling autonomous coding agents in containerized and cloud-native environments. Developed as a Go-based platform with Kubernetes-native custom resource definitions (CRDs), it provides the foundational infrastructure needed to run untrusted agent workloads safely across sandboxed compute nodes. In day-to-day operation, Agenttier manages the agent lifecycle through API controllers, CLI tooling, and an integrated Web UI. It allows developers to deploy agents built with frameworks like LangGraph into isolated execution sandboxes using pre-configured Docker images. Because it acts as an orchestration tier rather than a standalone prompt client, it is model-agnostic and relies on the underlying agent frameworks and Python SDK bindings to interface with foundation model providers. What distinguishes Agenttier from single-user CLI coding tools is its focus on platform-level infrastructure. It incorporates Helm charts, Terraform configurations, and Kubernetes controller primitives (`api/v1alpha1`) to handle agent provisioning, networking, execution sandboxes, and lifecycle telemetry for multi-agent workloads at scale. Agenttier is distributed as an open-source project on GitHub under active development, providing infrastructure templates for engineering teams looking to host their own self-managed AI agent backend.
Tags: agent harness, ai agent, cli, coding agent, developer tools, OpenSource
api/v1alpha1) and Go controllers. - Containerized Agent Environments — Runs code generation and execution tasks within isolated Docker runtime images. - Python SDK Integration — Offers programmatic SDKs to bridge agentic frameworks such as LangGraph with Agenttier's orchestration platform. - Infrastructure-as-Code Support — Includes Helm charts and Terraform modules for automated deployment onto cloud Kubernetes clusters. - Integrated Management Web UI — Provides a web-based dashboard alongside CLI tools for monitoring active agent sessions and sandboxes.Agenttier is aimed at platform engineers, DevOps teams, and AI developers who need infrastructure to host, isolate, and scale autonomous coding agents on Kubernetes or containerized backends. It is not intended for end users looking for a standalone desktop CLI assistant out of the box without underlying infrastructure setup.
bash git clone https://github.com/agenttier/agenttier.git cd agenttier go build -o bin/agenttier ./cmd/... To deploy the platform onto a Kubernetes cluster using Helm: bash helm install agenttier ./helm/agenttier To install the Python SDK for agent development: bash pip install ./python-sdkbash # Verify cluster connectivity and apply CRDs kubectl apply -f config/crd/ # Run the agenttier controller locally ./bin/agenttier runpython from agenttier import AgentRunner runner = AgentRunner(endpoint="http://localhost:8080") session = runner.create_session(image="agenttier/general-coding:latest") result = session.execute_task("Refactor pkg/api to support context timeouts") print(result.status)Part of CLI Coding Agents
Categories: AI Coding Agents, Code Assistants, Developer Tools