Local-first agent platform that runs on your machine and is controlled via WhatsApp, Telegram, Slack, or Discord. Connects to your own LLM (Claude, GPT, DeepSeek). Over 100 pre-built skills available through the ClawHub marketplace. Includes a proactive heartbeat scheduler for autonomous task execution. Open source under MIT license.

Goal-driven agent development framework that generates node graphs from natural language objectives, self-heals on failure, and redeploys automatically. A hybrid approach: define the goal conversationally and the framework builds and wires the agent for you. Less manual wiring than LangGraph, closer to an agent builder than a ready-to-run platform.

Autonomous agent acquired by Meta. Provide a goal and it browses, codes, and deploys. Recently expanded to a desktop app with local file access and OS-level agent capabilities — moving beyond cloud-only toward an operating-system layer.

No-code automation agent platform with over 7,000 integrations and pre-built templates for common workflows such as email triage, scheduling, and CRM updates. The closest enterprise-safe alternative to OpenClaw.

Visual AI automation platform for building and deploying multi-agent workflows without deep technical expertise. Agents run on schedules or event triggers across 50+ integrations. Includes Gumstack — a governance layer that monitors and controls AI activity organisation-wide, not just within Gumloop.

Enterprise ready-to-run agent platform with pre-built connectors and governance baked in. The safe enterprise pick for organisations already in the Microsoft ecosystem.

Pre-built agents for enterprise workflows with a strong compliance posture. Relevant for regulated industries requiring HIPAA or financial services guardrails.

Autonomous coding agent. Provide a GitHub issue and it writes code, runs tests, and opens a pull request. Focused specifically on software engineering workflows.

Autonomous coding agent that runs directly inside VS Code. Reads and writes files, runs terminal commands, and browses the web — all within the IDE. Open source with support for any API-compatible model via OpenRouter, Anthropic, OpenAI, or local Ollama.

Multi-agent coding platform for VS Code. Runs a coordinated team of specialised AI agents in the editor — orchestrator, architect, coder, and reviewer roles working in parallel. Forked from Cline with a focus on multi-agent coordination rather than single-agent autonomy.

All-in-one AI coding agent for VS Code with support for 100+ model providers through a single interface. Positions itself as the broadest model coverage in any IDE agent — swap providers without reconfiguring workflows.

Agentic coding assistant and builder platform with a large developer user base. Provides code search, generation, and autonomous agent capabilities targeted at developers who want a fast, integrated build experience without switching tools.

High-performance multi-agent runtime with session management, memory, knowledge, and MCP support. Fast startup times. Optional managed platform available for deployment and monitoring.

Type-safe agent framework built by the Pydantic team. Structured input/output with strong validation. Ideal when parameter correctness matters, such as insurance claims or financial data pipelines.

Ultra-minimal code-agents from Hugging Face. The model writes and executes Python to achieve goals. Minimal abstraction, maximum control. Best suited for open-source and local model deployments.

TypeScript-first agent framework. For teams that are Node/TypeScript heavy and want agent tooling that fits naturally into their existing stack without context-switching to Python.

Minimalist agent framework: agents, handoffs, guardrails, and tracing. Model-agnostic despite the name. Opinionated design that prioritises simplicity over flexibility.

Natural language analytics platform that lets business users ask questions in plain English and get instant insights from their data.