Shape agent capabilities into your workflow.

ChatDev 2.0 expands the original software-development agent system into a zero-code multi-agent orchestration platform. Users assemble nodes, tools, and models through configuration to run workflows such as document summarization, information collection, content publishing, or data visualization. A Python SDK and modular backend also let developers add new nodes, model providers, and tools.

Tabby offers a self-hosted alternative to cloud coding assistants and connects to developers through IDE extensions. Its OpenAPI interface can be integrated with existing infrastructure, while the server is self-contained and does not require a separate database or cloud service. It also supports consumer-grade GPUs for teams that want local control over models and code context.

ZeroClaw is a single-binary agent runtime that connects cloud or local model providers to channels such as Telegram, Discord, email, webhooks, and CLI. It supports tools, memory, scheduled or event-driven procedures, and a web dashboard. Supervised autonomy, workspace boundaries, and OS-level sandboxes gate riskier actions by default.

GitHub's official MCP Server connects compatible AI clients to repositories, code, issues, pull requests, Actions, and security-related data. You can use GitHub's hosted endpoint or run the server through Docker or a local binary, then limit the exposed surface with toolsets or individual tools. Available write operations depend on the permissions granted to the OAuth session or token.

Langfuse helps teams instrument LLM applications so they can inspect traces, sessions, retrieval steps, prompts, scores, and evaluations in one place. It also supports prompt versioning, datasets, playground experiments, and integrations with common SDKs and frameworks. You can use the hosted service or self-host with Docker Compose or Kubernetes. Self-hosted instances send basic usage telemetry by default, but it excludes raw traces and can be disabled.

Onyx is a self-hosted AI workspace for teams that want to use different LLMs through one interface and ground answers in organizational knowledge. It can index application data through more than 50 connectors or connect agents to external systems through MCP, with additional support for research workflows and code execution. Lite mode keeps a smaller chat-and-agents stack, while the standard deployment adds vector and keyword indexes plus background synchronization workers for a fuller knowledge workflow.

Scientific Agent Skills packages 148 reusable skills and access paths to more than 100 scientific databases across biology, chemistry, medicine, drug discovery, materials science, and scientific computing. It follows the open Agent Skills standard and can be installed in Cursor, Claude Code, Codex, and other compatible agents for multi-step research workflows.

NanoClaw hosts personal AI agents in separate Linux containers and connects them to channels such as Telegram, Discord, and WhatsApp, with per-agent workspaces, memory, and scheduled tasks. Channels and alternative model providers are added on demand, while customization is primarily code-based; setup requires Docker, Node, pnpm, and Claude Code.

PicoClaw packages a personal AI assistant as a portable Go binary for x86, ARM, RISC-V, MIPS, and other architectures, with a reported core memory footprint below 10 MB. It can connect to many model providers and chat channels, manage MCP server configurations, schedule jobs, and spawn asynchronous sub-agents. Most hosted model choices still require their corresponding credentials.

Composio creates a separate session for each user and gives an agent runtime access to app authentication, tool discovery, and execution without loading hundreds of definitions into context. It offers TypeScript and Python SDKs, a CLI, framework adapters, and hosted MCP endpoints for more than 1,000 app toolkits. Setup requires a Composio API key and the relevant connected user accounts.

Equip coding agents with Vercel-maintained guidance for auditing deployment cost, performance, caching, and reliability; optimizing React, Next.js, and React Native code; reviewing web accessibility and UX; enforcing documentation standards; and creating claimable Vercel deployments. Once installed, the relevant skill becomes available when matching tasks are detected.

Semantic Kernel is a model-agnostic SDK for building, orchestrating, and deploying AI agents and multi-agent systems across Python, .NET, and Java. It supports multiple model providers, plugins, structured workflows, and vector-database integrations. The project now identifies Microsoft Agent Framework as its successor and provides a migration guide, which is relevant when maintaining or starting an application.

GPT Researcher plans subquestions, gathers evidence in parallel from the web or local documents, tracks sources, and assembles the findings into long reports with citations and export options. Its model, retriever, and MCP data sources can be customized, but typical setups require provider and search API keys. The project is experimental and explicitly does not present its output as academic advice or a recommendation for research papers.

Cognee turns ingested data into a self-hosted knowledge graph that agents can remember and query across sessions. Developers use its Python API or CLI to remember, recall, forget, and improve information, combining semantic retrieval with graph relationships. It can run locally, in Docker, through Cognee Cloud, or on a consolidated PostgreSQL memory layer; an LLM provider key is normally required for local setup.

OpenAI Agents SDK lets Python developers define agents with instructions, tools, guardrails, and delegation between specialized workers. It includes session history, human approval points, tracing, sandboxed long-running agents, and realtime voice support. The framework can use OpenAI APIs or other supported model providers and requires Python 3.10 or newer.

Vibe Kanban has announced that it is sunsetting. Its existing workflow organizes work as prioritized kanban issues and creates agent workspaces with a branch, terminal, and development server. The interface supports diff review, inline comments, application previews, and switching among coding agents such as Claude Code, Codex, and Gemini CLI.

MLflow records and manages the development and operation of LLM applications, AI agents, and machine-learning models. Teams can start a server, enable automatic tracing, inspect calls and metrics in the UI, and use evaluation, prompt optimization, experiment tracking, model registry, and deployment workflows to improve their systems. It can run locally, on private infrastructure, or in cloud environments, making it useful when generative AI and traditional model lifecycles need a shared operational layer.

Crush is a terminal coding assistant that can switch models during a session while preserving context and keeping multiple project sessions. It uses language servers for additional code context and connects MCP servers over stdio, HTTP, or SSE. It supports many OpenAI- and Anthropic-compatible providers as well as local models; users need to configure a working provider, credentials, or local endpoint before sending model requests.