Coding & DevOps harnesses for AI agents
68 open-source Coding & DevOps harnesses an AI agent can use — MCP servers, SDKs, and adapters. Browse them on Loadbay. An agent can search these over Loadbay's MCP:
claude mcp add --transport http loadbay https://loadbay.xyz/api/mcp
→ Best Coding & DevOps harnesses (top picks, ranked)
- openclaw — Self-hosted personal AI assistant and coding-agent framework — a Claude-Code-style agent that runs its tools inside Docker sandboxes on any OS and plugs into multiple messaging channels. The lobster way.
- AutoGPT — Platform for building, deploying, and running continuous autonomous AI agents that automate complex workflows.
- opencode — Open-source AI coding agent that runs in the terminal and works with many model providers.
- langchain — Framework for building LLM applications and agents with composable chains, tools, and integrations across many providers.
- claude-code — Anthropic's agentic coding tool that lives in your terminal, understands your codebase, and runs tasks and git workflows in natural language.
- gemini-cli — Open-source AI agent that brings Google Gemini into your terminal for coding and automation.
- Codex — OpenAI's lightweight coding agent that lives in your terminal and edits your codebase.
- OpenHands — Autonomous AI software-engineering platform that writes code, runs commands, and browses the web in a sandboxed environment.
- MetaGPT — Multi-agent framework that assigns roles to GPTs to form a collaborative software company for natural-language programming.
- cline — Autonomous coding agent available as an SDK, IDE extension, or CLI assistant that can edit files and run commands.
- autogen — Microsoft programming framework for agentic AI enabling multi-agent conversations and collaborative task solving.
- Context7 — MCP server that injects up-to-date, version-specific library docs and examples into AI coding assistants to cut hallucinated APIs.
- OpenManus — Open-source framework for building general-purpose autonomous agents without an invite code or waitlist.
- gpt-engineer — CLI platform that generates entire codebases from a natural-language prompt, an early precursor to modern codegen tools.
- crewAI — Framework for orchestrating role-playing autonomous AI agents that collaborate as crews to tackle complex tasks.
- goose — Extensible open-source AI agent from Block that installs, executes, edits, and tests code with any LLM via MCP.
- aider — AI pair-programming agent in the terminal that edits code across your git repo with the LLM of your choice.
- agno — Full-stack framework (formerly phidata) for building, running, and managing agent platforms with memory, tools, and reasoning.
- dspy — Framework for programming language models with declarative modules and automatic prompt optimization instead of manual prompting.
- langgraph — Library for building stateful, resilient multi-agent applications as graphs with cycles, persistence, and human-in-the-loop control.
- continue — Open-source coding agent and IDE extension for building custom AI assistants inside VS Code and JetBrains.
- ChatDev — Virtual software company powered by LLM-driven multi-agent collaboration that builds software from a prompt.
- github-mcp-server — GitHub's official MCP server — let an agent read repos, manage issues and PRs, and search code through one standard tool surface.
- semantic-kernel — Microsoft SDK for integrating LLMs into apps with plugins, planners, and memory across C#, Python, and Java.
- smolagents — Barebones Hugging Face library for building agents that think and act by writing and executing code.
- openai-agents-python — Lightweight framework for building multi-agent workflows with handoffs, guardrails, and tracing from OpenAI.
- agentscope — Framework for building observable, trustworthy multi-agent applications with message-passing and workflow orchestration.
- Serena — MCP toolkit giving coding agents semantic, symbol-level code retrieval and editing across 40+ languages via language servers.
- qwen-code — Open-source AI coding agent that lives in your terminal, optimized for Qwen models.
- mastra — TypeScript framework for building AI apps and agents with workflows, memory, RAG, and evals.
- ai — TypeScript toolkit from Vercel for building AI apps and agents with streaming UI and many model providers.
- babyagi — Experimental framework for a self-building autonomous agent that generates and executes tasks toward an objective.
- swarm — Educational framework from OpenAI exploring lightweight, ergonomic multi-agent orchestration with routines and handoffs.
- Kilo Code — All-in-one agentic coding platform as a VS Code/JetBrains extension and CLI with Code, Plan, Ask, Debug, and Review modes; works with 500+ models.
- adk-python — Google's code-first Python toolkit for building, evaluating, and deploying sophisticated AI agents.
- suna — Open-source generalist AI agent platform that acts as a company command center across tasks and tools.
- SWE-agent — Autonomous agent that takes a GitHub issue and tries to fix it automatically using your language model of choice.
- pydantic-ai — Agent framework that brings type-safe Pydantic validation to building production-grade LLM applications.
- SuperAGI — Dev-first open-source autonomous agent framework for building, managing, and running useful autonomous agents.
- camel — Multi-agent framework for studying the scaling laws and communicative behaviors of autonomous cooperating agents.
- Qwen-Agent — Agent framework on Qwen models with function calling, MCP, a code interpreter, and RAG for building LLM applications.
- E2B — Open-source infrastructure to run AI-generated code securely in isolated cloud sandboxes, with Python and JS/TS SDKs.
- Trae Agent — ByteDance LLM-based CLI agent for general software-engineering tasks across multiple model providers and tool-driven workflows.
- agent-framework — Microsoft framework for building, orchestrating, and deploying AI agents and multi-agent workflows in Python and .NET.
- GitHub Copilot CLI — Brings GitHub Copilot's coding agent to the terminal for natural-language development and repo operations.
- AWS MCP Servers — AWS's open collection of MCP servers, so an agent can build on and operate AWS services.
- GitMCP — A remote MCP server that points an agent at any GitHub repo's real code and docs, to stop it hallucinating APIs.
- agent-squad — AWS Labs framework for managing multiple AI agents and routing complex conversations across them.
- swarms — Multi-agent orchestration framework for building and coordinating production-grade swarms of collaborating agents.
- strands-agents — Open-source SDK for building and controlling production AI agent harnesses across any model or cloud.
- atomic-agents — Lightweight framework for building AI agents from small, composable, schema-driven atomic building blocks.
- mcp-atlassian — MCP server connecting AI agents to Atlassian Jira and Confluence for issue tracking and documentation workflows.
- mcp-server-cloudflare — Official Cloudflare MCP servers for managing Workers, DNS, logs, and other services from an AI assistant.
- beeai-framework — Framework for building production-ready AI agents and multi-agent workflows in both Python and TypeScript.
- Grafana MCP — Official Grafana MCP server giving AI assistants access to dashboards, datasources, incidents, and observability data.
- supabase-mcp — Official Supabase MCP server connecting AI assistants to projects to run SQL, manage tables, and deploy functions.
- kubernetes-mcp-server — An MCP server for Kubernetes and OpenShift — inspect resources and run cluster operations from an agent.
- GitLab MCP Server — MCP server letting AI clients manage GitLab projects, merge requests, issues, pipelines, wikis, and releases.
- terraform-mcp-server — Official HashiCorp MCP server integrating the Terraform ecosystem to automate infrastructure-as-code workflows for AI agents.
- codedb — Code intelligence MCP toolset for AI agents with fast outline, symbol search, read, edit, deps, and remote GitHub repo queries.
- code-index-mcp — MCP server that helps LLMs index, search, and analyze code repositories with minimal setup.
- atlassian-mcp-server — Official remote MCP server from Atlassian that securely connects Jira and Confluence to LLMs, IDEs, and agent platforms.
- sentry-mcp — Official Sentry MCP server letting LLMs query and analyze application errors and issues from Sentry.
- mcp-server-docker — MCP server that lets AI agents manage Docker containers, images, volumes, and networks through natural language.
- sandboxed.sh — Runtime for autonomous on-chain AI agents with isolated sandboxes, encrypted secrets, and OKX read-only security checks.
- OpenHands CLI — Lightweight terminal CLI for the OpenHands coding agent, runnable in a shell, IDE, CI pipeline, or cloud sandbox.
- mcp-server-datadog — MCP server for Datadog giving AI agents access to monitors, metrics, logs, and incidents via the Datadog API.
- mcp-server-circleci — Official CircleCI MCP server bridging CI/CD pipelines and AI agents to inspect builds and fix failing tests.