Async Agents

Autonomous & Async Agents

Agents that operate independently on tasks, often outside your local editor.

23 tools · 23 with full reviews · 11 free or open source · Updated Sep 25, 2026

Google Jules is an asynchronous, autonomous AI coding agent that integrates with GitHub to perform tasks like bug fixing, dependency management, and feature implementation. It operates in secure cloud virtual machines, allowing users to queue tasks and receive pull requests upon completion.

OpenHands Freemium

All Hands OpenHands is an open-source, model-agnostic autonomous coding agent designed to execute full engineering tasks like bug fixes, PR reviews, and legacy code migrations within isolated Docker or Kubernetes sandboxes. It features an Agent Canvas control center that allows users to manage multiple agents, automate workflows via webhooks, and integrate with platforms like GitHub, GitLab, Slack, and Jira.

SWE-agent Open Source

Princeton NLP SWE-agent is a specialized autonomous agent designed to read GitHub issues, navigate repositories, and generate tested pull requests. It utilizes a purpose-built Agent-Computer Interface (ACI) to interact with codebases, effectively turning an LLM into an autonomous software engineer for bug fixing and maintenance tasks.

Sweep Sweep is an agentic layer for enterprise systems that connects, interprets, and governs metadata across platforms like Salesforce, Snowflake, and HubSpot. It allows teams to visualize dependencies, automate operations, and execute governed changes through AI agents.

Ellipsis Ellipsis is a managed cloud platform that enables teams to deploy autonomous coding agents defined as version-controlled YAML files within their repositories. It provides isolated, ephemeral sandboxes for tasks such as automated code review, bug fixing, and workflow automation, supported by full observability and hard budget caps.

Cosine Genie is a specialized autonomous AI software engineering agent developed by Cosine, designed to handle complex, multi-step engineering workflows by autonomously planning, writing, and testing code. The platform has pivoted toward the Lumen specialist model family, focusing on maintainable, production-quality output rather than general-purpose assistant tasks.

Open-yak Open Source

Open-yak OpenYak is a local-first, open-source AI desktop agent that integrates directly with your local files, native applications, and web browsers. It acts as an agentic workbench that manages long-running workflows, tool execution, and artifact generation while keeping your data and state on your own machine.

DeepAgents JS Open Source

LangChain DeepAgents JS is an opinionated, ready-to-run agent harness built on top of LangGraph that provides built-in capabilities like task planning, file system access, and sub-agent delegation. It is designed to be highly customizable, allowing developers to swap models, tools, and middleware while maintaining production-ready features like streaming and persistence.

Orkas-AI Orkas is an open-source, local-first desktop AI client that allows you to command a team of specialist agents via chat. It handles complex, multi-step work across research, coding, documentation, and operations by coordinating autonomous sub-agents that use local files and connected tools.

Kent C. Dodds Kody is a personal assistant platform that serves as a durable home for your AI agent's memory, secrets, and custom automations. By connecting your existing MCP-capable agents like ChatGPT, Claude, or Cursor, you enable them to execute code, save packages, and run scheduled tasks in the cloud independent of your active chat sessions.

Omnigent Omnigent is a meta-harness that provides a unified orchestration, collaboration, and policy layer for AI coding agents. It allows developers to swap or combine various agent frameworks like Claude Code, Codex, and Pi without rewriting code, while enforcing security and cost policies.

Monica Manus is an autonomous AI agent platform that plans and executes multi-step tasks by operating within a sandboxed virtual cloud computer. Instead of a standard chatbot interface, it takes a high-level goal and independently uses web browsers, terminal code execution, and file tools to deliver finished artifacts like research reports, slide decks, or working web applications.

LoopTroop AI LoopTroop is a local-first GUI orchestrator designed for complex, long-running AI software delivery that prioritizes correctness over raw speed. It breaks down large coding tasks into atomic, verifiable units called beads and utilizes multi-model councils and automated retry loops to prevent context rot and hallucination.

Meta Muse is a personal AI agent designed to execute multi-step tasks autonomously, such as shopping, email management, and travel booking, rather than simply answering questions. It operates within a secure, dedicated virtual machine and requires explicit user approval for sensitive actions like financial transactions.

OtoDock Freemium

OtoDock OtoDock is an open-source, self-hosted agent platform that orchestrates Claude Code and Codex CLI engines into collaborative, multi-tenant agent teams. It runs on your own infrastructure and manages state, permissions, and tools through a centralized dashboard.

firstmate Open Source

kunchenguid firstmate is an "agent distro": a portable directory of AGENTS.md instructions, bundled skills, helper scripts, policies, and state conventions that turns any terminal coding agent into a first mate that commands a crew. You talk only to the first mate; it clones your projects, spawns autonomous crewmates in a visible session backend (tmux by default, or herdr, zellij, cmux, and Orca), gives each one a disposable git worktree, supervises them with an event-driven bash watcher that wakes the primary session only when a decision is needed, and hands back finished pull requests, approved local merges, or standalone scout reports. Projects declare a mode (`no-mistakes`, `direct-PR`, `local-only`, optional `+yolo`), the first mate is read-only over your repos outside guarded operations, all state lives on disk so sessions are restart-proof, and larger fleets can opt into persistent secondmates running from isolated homes locally or over SSH.

OpenResearch Open Source

alphaXiv OpenResearch from alphaXiv is a local-first workspace that turns Claude Code, Codex, OpenCode, or Cursor into research agents that review literature, form hypotheses, edit code, launch experiments, inspect results, and produce research artifacts. `orx up` starts a local dashboard on 127.0.0.1:4791 backed by SQLite; each research direction gets its own agent session and isolated git worktree, and variants are tracked in a git-native experiment tree where every run receives an immutable archive of its recorded commit with logs, diffs, and artifacts tied to it. In autoresearch mode, agents run the full propose → change → launch → inspect → decide loop autonomously, in parallel. The same committed snapshot can execute locally, over SSH (`orx up --remote user@host`), or on Slurm, Kubernetes, Ray, Hugging Face Jobs, Modal, Tinker, and managed compute, and `orx install-skills` adds an OpenResearch skill to supported coding agents along with `orx discover` and `orx paper` for literature lookup.

Hermes Agent Freemium

Nous Research Hermes Agent is a self-improving, autonomous AI agent that features a closed learning loop, allowing it to extract, refine, and reuse skills from past task experiences. It supports persistent memory across sessions and acts as a multi-platform gateway for over 20 messaging services including Telegram, Discord, and Slack.

Coder Coder is a self-hosted platform for cloud development environments and AI coding agents. Workspaces are declared as Terraform templates that provision EC2 VMs, Kubernetes pods, Docker containers, or other infrastructure, are reached through a secure WireGuard tunnel from VS Code, JetBrains, or the terminal, and shut down automatically when idle. Coder Agents runs a native AI coding agent whose loop executes in the Coder control plane on your own infrastructure, so no LLM credentials live inside workspaces, every action carries the user's identity, and you can bring any model (Anthropic, OpenAI, Google, Bedrock, or self-hosted). The AI Gateway centralizes authentication, auditing, and cost tracking for AI tooling, and Coder Registry modules run Claude Code, Codex, and OpenCode isolated inside workspaces.

Google AX (Agent Executor) is Google's open-source distributed harness runtime that dynamically provisions isolated, suspendable/resumable environments to execute agents and harnesses at scale across a cluster. It provides a Kubernetes-like, declarative orchestration layer for running large numbers of autonomous agent workloads.

GitHub Agentic Workflows Open Source

GitHub GitHub Agentic Workflows (gh-aw) lets developers define AI-powered repository automation in Markdown with YAML frontmatter and compile it into GitHub Actions workflows. A single `gh aw` command turns a natural-language workflow into a locked-down `.lock.yml` that runs coding agents on issues, pull requests, schedules, or commands.

Strand Agents Open Source

Strand Strands Agents is an open-source, model-driven SDK for building and running production AI agents in Python and TypeScript with just a few lines of code. It pairs any model with tools via an agent loop, and the newer Strands harness offers a fully assembled, deployable general-purpose agent.

Paperclip Open Source

Paperclip Labs Paperclip is an open-source control plane for running a team of AI agents like a company: if OpenClaw is an employee, Paperclip is the company. You define a goal, hire agents into an org chart with roles, titles and reporting lines (any bot, any provider — OpenClaw, Claude Code, Codex, Cursor, bash or HTTP, anything that can receive a heartbeat), approve the strategy, set budgets and monitor from a task-manager-style dashboard or your phone. Agents wake on scheduled heartbeats, check out ticketed tasks that carry full goal ancestry, delegate up and down the org chart, and resume persistent task context across runs. Task checkout and budget enforcement are atomic so there is no double work or runaway spend; approval gates, revisioned config with rollback, full tool-call tracing and an immutable audit log provide governance; and every entity is company-scoped so one deployment can run many isolated organizations with portable, secret-scrubbed import/export templates. A Skill Studio, shared org-wide skills, evals and performance reviews round out the agent-training pillar.

FAQ

What are the best Async Agents tools?

The most complete Async Agents listings on ai.dosa.dev include Jules, OpenHands, SWE-agent. Every tool has a detail page with pricing, key features, and a verdict - see the full list of 23 tools below.

Are Async Agents tools free?

11 of the 23 Async Agents tools listed here are free or open source. Others offer freemium or paid plans - check each tool's detail page for current pricing.

How is this list curated?

ai.dosa.dev is manually curated from community submissions and editorial review. Entries are enriched with pricing, features, and reviewer notes, and the directory is updated continuously.