The index · everything in the almanac
Search
Look up a team, a repository, a gathering or a paper, by keyword or by meaning.
No matches for “”
The index · everything in the almanac
Look up a team, a repository, a gathering or a paper, by keyword or by meaning.
No matches for “”
No matches for “”
Layer
Startupkeyword match
The open platform for cloud coding agents. · San Francisco, CA
OpenHands (formerly All Hands AI, born from the OpenDevin project) builds the major open-source coding-agent platform — agent canvas, sandboxed execution, SDK and an enterprise control plane — now rebranded under the project's name.
Open sourcekeyword match
Browser Use/browser-use · Python
Browser Use is an open-source Python library for browser agents. Version 1.0 (November 2026) replaced the 0.x DOM-extraction agent: the model now writes Python cells in one persistent REPL that calls browser-harness helpers and the Chrome DevTools Protocol directly. It runs on any LLM, locally or on Browser Use Cloud, and also ships a CLI and MCP server.
Open sourcekeyword match
OpenHands (formerly All Hands AI)/OpenHands · TypeScript
OpenHands (formerly All Hands AI) is an open-source coding agent platform — agent canvas, sandboxed execution, SDK and an enterprise control plane — with 500+ contributors and millions of downloads.
Open sourcekeyword match
Nous Research/hermes-agent · TypeScript
Nous Research's self-improving personal agent: a learning loop turns experience into reusable skills and builds a model of the user across sessions. It runs from a terminal UI, a desktop app or chat apps such as Telegram, Discord and Slack, all attached to one gateway that owns every session, on anything from a $5 VPS to serverless backends.
Startupkeyword match
Open-source sandboxes for running AI-generated code. · Prague, Czechia
E2B gives agents secure, isolated cloud micro-VMs to execute code and commands — the execution runtime layer of the harness, used by many coding and data-analysis agents.
Startupkeyword match
Elastic sandboxes for running AI-generated code.
Daytona provides secure, OCI-compatible sandbox 'computers' that spin up in under 90ms — with snapshots for persistent agent state. In 2026 it moved core development to a private codebase while continuing to run as a commercial platform.
Startupkeyword match
The programming language for agentic software. · United States
Agno (formerly Phidata) builds a high-performance open-source agent runtime — memory, knowledge, storage and monitoring built in — aiming to be the substrate beneath agentic applications rather than just a framework.
Startupkeyword match
Headless browser infrastructure built for AI agents. · United States
Browserbase runs fleets of headless browsers as an API — sessions, proxies, stealth, debugging — so agents can act on the web without operators managing browser infrastructure. Also builds the Stagehand SDK.
Startupkeyword match
The execution layer that 'parents' AI agents. · San Francisco, CA
Runta builds an agent execution layer with sandboxes, spending caps and audit trails — a 'parenting' layer for agents that directly addresses the containment failures exposed by prompt-injection incidents. a16z called it 'core systems software rebuilt for agents'; $20M seed at a $100M+ valuation (Jul 2026).
Startupkeyword match
Trust and infrastructure for open-source ecosystems — now Mecatl. · Seattle, WA
Stacklok, founded by Kubernetes/Sigstore creator Craig McLuckie, builds trust tooling for open source. In 2026 it shipped Mecatl — an open-source, cloud-native agent harness that architecturally separates the agent loop from code execution, built to run identically on a laptop or Kubernetes.
Startupkeyword match
Open AI lab behind the Hermes models and Hermes Agent. · Remote-first
Nous Research, known for open-weight Hermes LLMs and decentralized training, ships hermes-agent — a self-improving, always-on personal agent that turns experience into reusable skills and keeps a persistent user model, lean enough to run on a $5 VPS.
Startupkeyword match
The sessionless, relentless 24/7 personal agent.
Comma is an always-on personal agent with no session boundaries — 'no full stop until it's done'. Its distributed harness, Salix, loads tiny eBPF programs into the cluster to watch events around the clock and wake the main agent loop only when something needs it.
Startupkeyword match
Open-source AI agents that work for you.
TinyHumans builds a family of open-source agents: OpenHuman, a local-first Rust harness with private lifelong memory; OpenCompany, an agent 'hive mind' that thinks as one team; and Medulla, a terminal that runs Claude Code, Codex and OpenCode side by side.
Startupkeyword match
Build, deploy and publish agent apps as MCPs.
Recurse makes it easy to build an agent app, deploy it, and publish it as an MCP server usable from Claude, Codex or any MCP client — hosting handled, started from your coding agent via a SKILL.md. A bridge between harness skills and deployable agent infrastructure.
Startupkeyword match
Multiplayer cloud workstations for coding agents and humans.
Amika's 'Rigs' are cloud VMs loaded with your dev setup and favorite coding agents — booted in seconds, steered from web, Slack, Linear, SSH, CLI or API, and shareable so teammates can inspect previews or take over an agent session. Open core.
Startupkeyword match
A multiplayer agent harness for work.
QM gives every person and shared room a durable, isolated agent workspace while keeping administration, permissions and background work centralized — docs cover models and harnesses, sandboxes, apps and skills, and Slack/web surfaces. Its open-source runtime (open-multi-agent) ships durable approvals and verifiable run records.
Startupkeyword match
The browser API for AI agents. · San Francisco, CA
Steel Dev provides a batteries-included open-source browser (steel-browser) plus a cloud Browser API for agents and apps — isolated Chrome sessions, stealth, residential proxies, CAPTCHA handling and persistent profiles. The open-source core in the hosted browser-infrastructure lane.
Startupkeyword match
Open drivers, cross-OS sandboxes and benchmarks for computer-use agents. · San Francisco, CA
Cua (YC Spring 2025) builds infrastructure for computer-use agents. Cua Spaces is a desktop app that gives agents full desktops: macOS VMs built locally, or Linux images, on the user's own machines or cloud. Cua Driver lets agents inspect and operate native apps and browsers on macOS, Windows and Linux through a CLI, MCP or typed SDKs. The stack also includes Lume (macOS VMs), the Cua SDK and CLI for local sandboxes, CUA-S1 small models for computer-use decisions, and Cua Bench for simulated tasks. The repo is MIT-licensed, while the Spaces app is source-available under FSL-1.1-MIT.
Startupkeyword match
Open-source durable runtime for building and deploying AI agents and workflows in TypeScript.
Trigger.dev is an open-source (Apache-2.0) platform for long-running AI agents and workflows in TypeScript. Tasks run without timeouts and get durability via checkpointing, retries, queues, concurrency controls, idempotency and versioned deploys, plus human-in-the-loop waitpoints, Realtime streaming of LLM output to frontends, cron schedules and per-run tracing and logs. Its chat agents are pitched as surviving refreshes, redeploys and crashes. It can be used as a managed cloud with usage-based pricing or self-hosted. In December 2025 the company announced a $16M Series A led by Standard Capital with participation from Y Combinator.
Startupkeyword match
TrueForge: open-source agent harness runtime layer.
TrueFoundry, which sells an enterprise AI gateway and MCP gateway, maintains TrueForge, an open-source (MIT) agent harness that runs the agent execution loop: model calls, MCP tools (with header auth or OAuth), git-backed SKILL.md skills, sandboxed code execution (Daytona today), tool approvals, context management and session state. TrueForge exposes this through a chat UI, an HTTP API with a TypeScript SDK and an embeddable UI SDK, and adds context engineering such as subagents, deferred tool loading, Code Mode and compaction. It runs locally (SQLite) or hosted (Postgres + Redis).
Startupkeyword match
Run a team of coding agents in parallel isolated workspaces, locally or in the cloud. · San Francisco, CA
Conductor creates isolated copies of a codebase, runs coding agents in each, and lets developers review, open PRs for and merge their work. It runs the first-party Claude Code, Codex, Cursor and OpenCode agents with the user's own subscriptions and keys. Conductor Cloud gives each agent a Firecracker microVM sandbox with repos and dependencies pre-installed, and workspaces are multiplayer. There are macOS and iOS apps and an API. Founded by Charlie Holtz and Jackson de Campos (YC S24), it raised a $22M Series A from Spark and Matrix in March 2026.
Startupkeyword match
Cloud platform that runs any coding agent in shareable sandboxed environments.
Tembo runs AI coding agents (Claude Code, Codex, Cursor, Amp, OpenCode, Pi and others) in isolated cloud environments with source code, terminal, files and a live desktop, turning issues and alerts from Sentry, Linear and Jira into pull requests, including coordinated multi-repo changes. Tembo Previews spin up frontends, APIs and databases behind tunnel links; agents reach databases and team tools via MCP. It offers per-task model routing, scheduled tasks, webhooks, a public API and TypeScript SDK, plus SSO, RBAC, audit trails and self-hosted/VPC deployment.
Startupkeyword match
Background software-engineering agents in governed, VPC-deployable cloud environments.
Ona, the company formerly known as Gitpod (Gitpod GmbH), runs AI engineering agents in full cloud development environments that can be deployed in the customer's VPC. Background agents take a task and return a pull request; automations trigger agent fleets from PRs, schedules or webhooks for work like CVE remediation and code migration. Runtime controls include audit trails, scoped credentials and kernel-level policy enforcement, and teams can run Ona, Codex and other agents under shared policies. OpenAI announced on June 11, 2026 that it is acquiring Ona, with the team joining Codex.
Startupkeyword match
Production-like environments with scoped access to real infrastructure for coding agents. · San Francisco, CA
Crafting provides secure agent infrastructure running in the customer's cloud: sandboxes that mirror production, giving agents scoped access to database snapshots, internal services, caches and queues, with network isolation and credentials injected only when needed and never exposed to the model. Crafting for Agents, generally available since March 2026, adds Kubernetes interception, multi-service orchestration, hot-swapping of services and multi-region failover, complementing existing CI/CD. San Francisco-based; raised a $5.5M seed led by Mischief with WndrCo (March 2026). CEO Sumeet Vaidya.
Startupkeyword match
Devbox sandboxes and agent infrastructure for AI coding agents.
Runloop Devboxes are VM-isolated sandboxes where AI agents build, run and test code, starting in seconds and supporting snapshot, suspend and resume for long-running work. Blueprints define team-shared images; network policies restrict egress, while Agent Gateways and an MCP Hub connect agents to LLM APIs and MCP servers without exposing real credentials. The platform also lists measurement for evaluating agent performance, multi-agent coordination, Reflex debugging/observability and VPC deployment on AWS, GCP or Azure, with Python and TypeScript SDKs and a CLI.
Startupkeyword match
Perpetual-standby microVM sandboxes for production agents, now part of Baseten. · San Francisco, CA
Blaxel (YC Spring 2025) provides compute, storage and networking for autonomous agents. Each sandbox is an isolated microVM that resumes in about 25ms with memory state intact and can stay on standby indefinitely when idle; it also offers batch jobs across sandboxes, MCP servers as first-class workloads, persistent volumes, egress controls and a model gateway. Founded in 2024 in San Francisco by Paul Sinai, Christophe Ploujoux and others, it was acquired by AI inference company Baseten in 2026 (terms undisclosed); YC lists it as acquired.
Startupkeyword match
Full Linux VMs for AI agents that boot fast, clone live and hibernate. · San Francisco, CA
Freestyle (YC S24) builds VMs for AI agents intended for tasks running hours to weeks. Its site lists fast VM startup, live cloning of running VMs including memory, pause and resume with state intact, default-deny networking with per-VM rules, live preview URLs with custom domains, WireGuard tunnels to private networks, nested virtualization for Docker, and a full kernel with FUSE and eBPF. YC describes VMs that can be snapshotted and hibernated. Founded in 2024 in San Francisco by Benjamin Swerdlow and Jacob Zwang.
Startupkeyword match
Moonshot AI's hosted, always-on OpenClaw deployment on kimi.com.
Kimi Claw is Moonshot AI's managed version of the OpenClaw personal-agent harness, announced in February 2026. It deploys an OpenClaw assistant to the cloud with one click to run 24/7, auto-configured with Kimi models and skills; Dataconomy reports it runs on Kimi K2.5 with 40GB of cloud storage and access to ClawHub's 5,000+ community skills. A 'Bring Your Own Claw' option links existing self-hosted OpenClaw setups, and the page also offers local desktop and Android deployment. The kimi.com page marks it as beta.
Startupkeyword match
Managed agent platform: pick the harness, model and tools; Ellipsis runs it
Ellipsis runs coding agents on its own infrastructure and handles VMs, permissions, budgets and logs. Agents are defined in YAML: trigger, harness and model, prompt, repos, hooks, compute and spend limits. Supported harnesses include Claude Code, Codex, OpenCode, Gemini CLI, GitHub Copilot and Cursor. Use cases are software factories, code review and background agents. The company is backed by Y Combinator.
Startupkeyword match
Google's agentic development platform
Google Antigravity is an agentic development platform with four parts. The App is a command center for running multiple local agents in parallel. The CLI is terminal-first, with background subagents. The IDE is agentic, with an agent manager and artifacts. The Python SDK is for prototyping custom agents on the Antigravity harness.
Startupkeyword match
Your developers have guardrails. Your AI doesn't.
Coder provides self-hosted AI development infrastructure that lets developers use any model or agent while giving enterprises governance and control. Coder Agents is a native coding agent that works with any model on infrastructure you control. Coder AI Governance centralizes model access, enforces policy-controlled environments and keeps audit trails. Coder Workspaces are standardized, self-hosted environments shared by developers and agents.
Startupkeyword match
Local validation for your agent's code
Chunk is a CircleCI tool that validates code written by Claude Code, Cursor or other agents before it is pushed to CI. It syncs local code to a remote Linux microVM sidecar and runs tests there as 'microbuilds' inside the agent loop, using hooks to give deterministic feedback. Sidecars can run integration suites with services like Postgres and Redis. It is free on every CircleCI plan.
Startupkeyword match
Open-source agent that lives on your machines 24/7 and keeps your apps running
Stakpak is an open-source DevOps agent written in Rust that runs on your own machines around the clock to keep apps running, contacting a human only when needed. It is positioned as offering the upside of a PaaS without lock-in. Its website states that Stakpak is joining Vercel.
Startupkeyword match
Build websites, apps and prototypes by chatting with AI
Bolt is a chat-based AI tool for building websites, apps and prototypes, with Bolt Cloud providing hosting, databases, authentication and integrations. Its Bolt Agent (Standard and Max tiers) automatically picks a model per task and tests, refactors and iterates to reduce errors.
Startupkeyword match
Cloud browsers, sandboxes and web data APIs for AI agents
Hyperbrowser hosts cloud Chrome sessions that can be controlled with Playwright or Puppeteer, with persistent logins and live inspection. It also offers isolated sandboxes with snapshots and a web API for fetching, crawling and search that returns Markdown, HTML, screenshots or JSON matching a schema. Agent tasks can be run with HyperAgent, Browser Use or computer-use models, and the service ships Node.js and Python SDKs, a CLI and MCP.
Startupkeyword match
Cloud-hosted browser infrastructure for computer-use and browser agents
Anchor describes itself as a cloud-hosted browser that lets AI agents use the web the way a person would. Features include Anchor Chromium (a stealth Chromium fork), OmniConnect for login handling, a built-in VPN and a Web Action Cache that turns planned tasks into deterministic code. It can be deployed in the customer's own cloud and is reachable through an SDK or MCP.
Startupkeyword match
Browser infrastructure for web agents and automations
Kernel provides isolated cloud Chromium sessions controlled through an API, SDK or CLI, using Playwright, CDP, WebDriver BiDi or computer use. It supports persistent profiles, managed auth that keeps credentials and MFA codes out of the agent conversation, and stealth mode with captcha solving and proxies. It also has live view, session replay, a hosted MCP server, and browser pools for latency-sensitive work.
Startupkeyword match
Managed headless browsers for automation and AI agents, over MCP, Puppeteer or Playwright
Browserless runs headless browsers in the cloud that Puppeteer, Playwright and Selenium connect to over WebSocket. It offers an MCP server with scraping, crawling, search and export tools, BrowserQL (a GraphQL interface with stealth options), and REST endpoints for screenshots, PDFs and scraping. It lists integrations with Browser Use, Stagehand, LangChain and Claude Code.
Startupkeyword match
Browser platform and framework for AI web agents
Notte provides cloud browser sessions with anti-detection, residential proxies and CAPTCHA solving, plus browser agents that navigate sites from natural-language tasks. It can deploy browser automations as serverless functions with cron scheduling, and offers personas and secret vaults so credentials stay away from the LLM. It has an MCP server, a CLI and an agent skill for coding agents. The site says it is backed by Y Combinator (W25).
Startupkeyword match
AI writes code. Riza runs it.
Riza is an API for running code written by an LLM in isolation from production systems and secrets, with per-run control over environment variables and network access. The site says execution starts in under 10ms with no sandbox startup. It offers Python, TypeScript and Go SDKs, lets agents write reusable tools, and can be self-hosted on Kubernetes.
Startupkeyword match
Sandboxes, persistent VMs and serverless GPUs for AI agents
Beam is a cloud platform for agents that offers persistent VMs, isolated sandboxes, inference endpoints and serverless GPUs. Sandboxes boot in seconds to run untrusted, agent-generated code, use custom images, and can fan out from one saved snapshot, for example for RL rollouts. The open-source runtime is beta9, and the site says the company is backed by Y Combinator.
Startupkeyword match
Persistent Linux computers for AI agents, with checkpoints and sleep/wake
Sprites is a hosted service from Fly.io that gives AI agents full Linux computers with persistent disks. It takes live whole-disk checkpoints and can restore from them, and idle Sprites sleep and wake when a request reaches their HTTPS URL. Connectors let agents call services like GitHub or Slack without the credential living inside the Sprite, and there are a CLI, a REST API and SDKs for JavaScript, Go, Python and Elixir.
Startupkeyword match
On-demand isolated Linux microVMs for running untrusted or AI-generated code
Vercel Sandbox runs arbitrary code in on-demand Firecracker microVMs, each with its own kernel, controlled through an SDK or CLI. Claude, Codex and OpenCode come preinstalled in the base image, so coding agents can run inside a sandbox. It has egress policies with credential injection on outbound requests, snapshots that can start parallel instances, sessions up to 24 hours on paid plans, and persistent Vercel Drives.
Startupkeyword match
Runtime platform for AI agents, their sandboxes and the software they build · London, England
Northflank runs agent workloads in isolated microVM sandboxes and can host coding agents such as Codex, Claude Code, OpenCode and Pi remotely with persistent storage. It also runs services, managed databases, jobs and GPUs, with preview environments per PR and pipelines. It runs on Northflank Cloud or in a customer's own cloud or bare metal, and the site states a registered office in London.
Startupkeyword match
AWS platform for building, connecting and running production AI agents with any framework or model
Amazon Bedrock AgentCore is AWS's platform for building, connecting and running production AI agents with any framework or model. It includes a harness for deploying agents and a Gateway that handles authentication and access control for tool calls to MCP servers, APIs and Lambda functions. It also has observability and evaluation tools for tracing agent behavior and testing variations against real traffic.
Startupkeyword match
AI Cloud in one line of code · San Francisco, CA
Langbase is a serverless platform for building, deploying and scaling AI agents, which are exposed as APIs (pipes) and can connect to 600+ LLMs, tools and MCP servers. It provides a managed memory/RAG service, and multi-step multi-agent workflows with retries, timeouts and step-level traces. It also offers Studio for versioning agents, plus tracing, cost tracking and evals. Its site also promotes Command Code, a coding agent.
Startupkeyword match
Just Use Postgres
DBOS makes workflows durable so applications and AI agents recover from failures, storing workflow state in an existing Postgres database with no separate orchestration server. DBOS Transact is an open-source durable execution library with Python, TypeScript, Go, Rust and Java versions, and integrates with Pydantic AI, LlamaIndex, the OpenAI Agents SDK and the Vercel AI SDK.
Startupkeyword match
The durable runtime for your AI agents
Golem runs AI agents and their tools in lightweight WebAssembly VMs and resumes them after crashes, restarts or long waits without repeating completed calls. The core runtime is open source on GitHub, written in Rust.
Startupkeyword match
The micro-VM for AI agents
BoxLite is a managed cloud that runs AI agents' apps in hardware-isolated micro-VMs, each with its own kernel and a public URL. It is built on an open-source runtime that can also run locally or on your own servers.
Startupkeyword match
Sandboxes with stateful twins of the services agents act on
Arga Labs provides sandboxes for testing and training AI agents, with stateful twins of third-party APIs, CLIs and MCPs such as Slack, GitHub and Gmail. Teams point their agents at the twins, seed them with data and review full traces of every call and side effect.
Startupkeyword match
Build locally, delegate to the cloud: runs Claude Code and Codex in cloud sandboxes and returns pull requests.
Twill runs coding agents such as Claude Code and Codex in cloud sandboxes, so a developer can hand off a task from their machine and get a pull request back ("Build locally. Delegate to the cloud."). It is a Y Combinator S25 company and launched on Hacker News in April 2026.
Startupkeyword match
Effortless cloud coding agents that carry your local sessions, memories and MCP servers to the cloud.
Hoplite runs coding agents in the cloud and ports a developer's local sessions, memories and MCP servers with them, so work can continue off the laptop. The Y Combinator S26 company pivoted into coding-agent infrastructure from an earlier retail-investing product and launched on Hacker News in August 2026.
Startupkeyword match
Isolated cloud sandboxes that host Hermes, OpenClaw, Claude Code, Codex, OpenCode, Pi and QM.
Agent37 hosts many agent harnesses — Hermes, OpenClaw, Claude Code, Codex, OpenCode, Pi and YC's QM — in isolated cloud sandboxes billed by the minute; its FAQ prices a 2 vCPU / 4 GB sandbox running all month at $4.76. Its founder says it "started as one of those OpenClaw wrapper companies back in February". Launch HN in October 2026 drew 163 points; Y Combinator F26.
Startupkeyword match
Database sandboxes for you and your coding agents, so agents test against a realistic Postgres.
Ardent gives coding agents Postgres database sandboxes. Its founders argue that without "a realistic sandbox at the DB layer for testing, they ship garbage that can take down production databases." The Y Combinator Spring 2026 company launched on Hacker News in May 2026.
Startupkeyword match
A rootless Linux sandbox with gVisor support for running agents safely.
Drop is an open-source, rootless Linux sandbox with gVisor support, built to run coding agents and other untrusted processes without giving them the host. It launched with a Show HN in September 2026 that drew 193 points and 63 comments.
Startupkeyword match
A full-stack cloud with forkable VM snapshots and global SQLite, deployable over SSH or Git.
Kedge is a cloud for deploying websites, functions and container images over SSH or Git; its Show HN in July 2026 (141 points) pitched forkable VM snapshots and global SQLite, features that suit agents which need disposable, branchable machines.
Startupkeyword match
The fastest persistent computer for agents; launched as "Vercel for Agents" connecting any LLM to any MCP tool.
Dedalus Labs launched on Hacker News in August 2025 as "Vercel for Agents", an SDK that connects any LLM to any MCP tool. Its site now positions the company as the "Fastest Persistent Computer for Agents", offering persistent VMs for agent workloads. Y Combinator S25.
Startupkeyword match
Launched as "Vercel for filesystem-based agents"; now rebuilds whole departments to run AI-first.
Terminal Use launched on Hacker News in March 2026 as "Vercel for filesystem-based agents", hosting agents that work through files and a shell. It has since pivoted: its site now offers "Whole departments, rebuilt to run AI-first" for construction, logistics, manufacturing and retail. Y Combinator W26.
Startupkeyword match
Launched as an API for computer-use agents on legacy apps; now builds living worlds and multi-agent environments.
Coasty launched on Hacker News in July 2026 with an API for computer-use agents that operate legacy desktop and web apps. Its site now reads "Living Worlds & Multi-Agent Environments", a pivot towards environments where many agents act. Y Combinator S26.
Startupkeyword match
Hosted OpenClaw: a personal OpenClaw agent running on its own cloud VM.
Klaus hosted OpenClaw for people who did not want to run it themselves, giving each user an OpenClaw agent on a dedicated virtual machine. It was one of the first OpenClaw hosting services, and one of the first in that wave to close, as the State of Harness October 2026 report noted.
Startupkeyword match
An agent environment from the lakeFS team: a versioned filesystem, credentials and tools for AI agents.
tilde.run, from the team behind lakeFS, was built on the idea that "AI agents are only as useful as the environment they can work in": a sandbox with a versioned filesystem, credentials, tools and execution for agents. Its lessons went into lakeFS Mount, an agent-native filesystem for production AI workloads.
Open sourcekeyword match
AWS/harness-sdk · Python
Strands is a model-driven agent SDK where the loop is directed by the model with typed tools, and batteries included for AWS. Its Python repo was renamed to harness-sdk in 2026 — AWS now talks the harness language natively, pitching it as a way to build an agent harness and control it end-to-end, with native MCP and multi-agent composition.
Open sourcekeyword match
Agno/agno · Python
Agno (formerly Phidata) markets itself as 'the programming language for agentic software' — a fast agent runtime with built-in memory, knowledge, storage and a monitoring UI, aiming to be the production substrate beneath frameworks.
Open sourcekeyword match
Cloudflare/agents · TypeScript
Cloudflare's Agents SDK builds agents on Durable Objects — stateful, always-on, running at the edge with built-in scheduling, WebSockets and AI bindings. The serverless answer to agent runtimes.
Open sourcekeyword match
E2B/E2B · Python
E2B runs untrusted LLM-generated code in secure Firecracker microVMs via a simple SDK — the code-interpreter substrate behind many agent harnesses.
Open sourcekeyword match
Daytona/daytona · Go
Daytona provides elastic, OCI-compatible sandbox 'computers' for agent code execution — spinning up in under 90ms with snapshots for persistent agent state. In June 2026 the open-source repo moved to a private codebase; the platform continues as a commercial product.
Open sourcekeyword match
Stacklok/mecatl · Go
Mecatl is Stacklok's open-source agent harness, architecturally separating the agent loop that reasons from the environment that executes code. The same backend runs on a laptop or a Kubernetes cluster — K8s-native from day one, not retrofitted — with TUI and web clients connecting to one engine. Early-stage but one of the few harnesses designed for platform teams.
Open sourcekeyword match
TinyHumans/openhuman · Rust
OpenHuman is TinyHumans' local-first, Rust-written agent harness: it builds a private memory of your life, runs agents and workflows for you, and does the deep research you'd never get to — online or offline, on hardware you own.
Open sourcekeyword match
QM (yc-software)/open-multi-agent · TypeScript
Open Multi-Agent is the open-source runtime behind QM: a self-hosted TypeScript agent runtime with durable approvals and verifiable run records — 'own it, approve it, audit it' — built for multiplayer agent work.
Open sourcekeyword match
Kubernetes SIGs/agent-sandbox · Go
Agent Sandbox is a Kubernetes SIG project: a Sandbox resource plus warm pools and claims for fast-start, isolated, stateful workloads. The self-hosted answer to hosted sandbox APIs — from the people who run half the world's clusters.
Open sourcekeyword match
Steel/steel-browser · TypeScript
Steel Browser is the open-source core of Steel's Browser API: cloud or self-hosted Chrome sessions with stealth, residential proxies, CAPTCHA solving and persistent profiles — the only open-source offering in the hosted browser-infrastructure lane.
Open sourcekeyword match
herdrdev/herdr · Rust
herdr is an Apache-2.0 Rust terminal workspace for coding agents such as Claude Code, Codex, Cursor, OpenCode and Grok. It does not wrap or replace those agents; it owns their terminals. A background server keeps sessions running after the client detaches or SSH drops. After a restart it restores layouts and can resume supported agent sessions. Each pane is marked working, blocked or idle. Agents can drive herdr through a CLI and socket API to spawn panes, prompt each other and wait on blocked peers. It has tmux-style keys, mouse support, multi-machine SSH and plugins.
Open sourcekeyword match
ZeroClaw Labs/zeroclaw · Rust
ZeroClaw is a personal-assistant agent runtime delivered as a single Rust binary that users configure and run on their own hardware. It talks to LLM providers including Anthropic, OpenAI, Ollama and about 20 others. It reaches users through 30+ channels such as Discord, Telegram, Matrix, email, voice, webhooks and a CLI. It acts through tools including the shell, a browser, HTTP and MCP servers. A web dashboard handles chat, memory, config and cron. Since v0.9 it can also sign in other people through your own OpenID Connect provider, each kept to a permission profile. Installs are documented for Linux, macOS, Windows, FreeBSD, NixOS and Docker.
Open sourcekeyword match
nanocoai/nanoclaw · TypeScript
NanoClaw is a personal AI assistant that runs each agent in its own Linux container, so agents get filesystem isolation instead of relying only on permission checks. Its author built it as a small alternative to OpenClaw, which the README describes as nearly half a million lines of code. NanoClaw is one process and a handful of files. It connects to WhatsApp, Telegram, Slack, Discord, Gmail and other channels, and supports memory and scheduled jobs. It runs on Anthropic's Agents SDK. Setup installs Docker, a credential gateway and the agent container. v2 can give each agent its own Slack bot identity, container and memory.
Open sourcekeyword match
NVIDIA/NemoClaw · TypeScript
NemoClaw is an open-source reference stack from NVIDIA for running supported AI agents more safely inside NVIDIA OpenShell sandboxes. It provides guided onboarding, managed inference, network policy, managed integrations, OpenShell-backed state persistence and lifecycle operations through the NemoClaw CLI. Supported agents are OpenClaw (the default), Hermes and LangChain Deep Agents Code; an express installer targets DGX and WSL hosts. The project describes itself as alpha, with maintainers reviewing issues and PRs on a best-effort basis.
Open sourcekeyword match
RightNow AI/openfang · Rust
OpenFang is an open-source Agent Operating System written in Rust and shipped as a single binary of about 32MB with a local dashboard. Its core concept is 'Hands': pre-built autonomous capability packages (manifest, system prompt, SKILL.md and guardrails such as approval gates) that run on schedules without prompting, e.g. Researcher and Lead. The site lists a WASM tool sandbox, Ed25519 manifest signing, a Merkle audit trail, SQLite plus vector memory, 40 channel adapters, and MCP and A2A support. The README marks it pre-1.0 (v0.5.10, April 2026) and warns of breaking changes.
Open sourcekeyword match
Google/ax · Go
Agent Executor (AX) is Google's open, declarative orchestrator for running autonomous agent workloads at scale. Users declare Tasks and Workspaces (git repos, files such as AGENTS.md, MCP servers, skills) plus Model and Gateway specs in YAML; AX sandboxes each task with CPU and memory limits, wires up its workspace, enforces network policies with credential injection, and checkpoints idle agents for sub-second resume. It runs on top of the Agent Substrate runtime and targets millions of tasks per cluster. The README warns it is in heavy development with major breaking changes likely before a stable release.
Open sourcekeyword match
NEAR AI/ironclaw · Rust
IronClaw is an open-source personal agent OS from NEAR AI, written in Rust, aimed at routine tasks such as email triage, meeting prep and deployment monitoring. Its security model runs untrusted tools in WebAssembly containers with capability-based permissions, keeps credentials in an encrypted vault injected only at the host boundary for approved endpoints, scans outbound traffic for leaked secrets and allowlists network access. Data is stored locally and encrypted; hosted instances can run in trusted execution environments on NEAR AI Cloud. It is model-agnostic (Anthropic, OpenAI, Ollama) and installs via a guided `ironclaw onboard` setup.
Open sourcekeyword match
Nasiko Labs/nasiko · Rust
Nasiko calls itself the 'OpenRuntime' for agents, coding harnesses, frameworks and tools. It discovers coding agents already running (Claude Code, Codex, Cursor, OpenCode) without changing their configuration, puts their spend into one schema broken down by harness and model, and can route their model calls to models the operator chooses while developers keep their existing harness workflows. Its site says it helps enterprises account for what their agents cost, can access and do across providers. Written in Rust, with a Docker-only quick start.
Open sourcekeyword match
smol machines/smolvm · Rust
smolvm provides branchable microVMs for AI agents: real VMs with their own kernel that boot in under a second, use elastic memory and run any OCI image on a laptop or own servers. Machines are declared in a Smolfile (image, resources, ports, mounts, network policy) and driven from code through a daemonless Machine SDK, locally or on smol cloud. Running machines can be checkpointed (RAM, CPU state, disks), rewound, or branched into copy-on-write children, and an arm64 Mac checkpoint can resume on arm64 Linux. Docs include SKILL.md files agents can follow.
Open sourcekeyword match
Microsoft/agent-governance-toolkit · Python
Microsoft's Agent Governance Toolkit is a public-preview library adding policy enforcement, agent identity, execution sandboxing and SRE-style reliability engineering to autonomous agents, installable with one pip install and framework-agnostic. It targets three questions: is an action allowed, which agent performed it, and can it be proven afterwards via tamper-evident decision records. The project argues prompt-level safety is not a control surface and claims coverage of 10/10 of the OWASP Agentic Top 10. Breaking changes are possible before GA.
Open sourcekeyword match
nolabs/nono · Rust
nono is an agent runtime security tool built on kernel-enforced sandboxing. It runs each tool in its own ephemeral sandbox with explicit file, network, environment and credential permissions, keeps API keys outside the agent's sandbox through proxy-based injection, filters network access by domain and API rule, verifies skills and instruction files with Sigstore, and keeps a tamper-evident audit log with file rollback. It is a single unprivileged binary for macOS, Linux and WSL2 and works with Claude Code, Codex, OpenClaw, Hermes and custom agents.
Open sourcekeyword match
ArcBox Labs/arcbox · Rust
ArcBox is an open-source container and VM runtime for macOS, written from scratch in Rust (its own VMM, VirtIO devices, filesystem sharing and networking) and positioned as an alternative to Docker Desktop and OrbStack. One daemon and CLI run four workload types: a drop-in Docker engine with native Kubernetes, disposable microVM sandboxes for AI agents and untrusted code, full Linux VMs, and throwaway macOS guests. Sandboxes run locally through Sandbox SDKs, with no cloud account required. It is in public beta.
Open sourcekeyword match
OneCLI/onecli · TypeScript
OneCLI is an open-source platform for running AI agents as a team. It started as a Rust credential vault for agents such as Hermes, OpenClaw and NanoClaw. Version 2 provisions one agent per employee from the company identity provider. Each agent runs in an isolated sandbox whose only exit is a gateway that injects credentials and enforces a single team policy. It supports deterministic human-in-the-loop approvals in chat, shared team-level connections and platform-kept memory. It can be self-hosted, including via Helm, or used through the hosted onecli.sh.
Open sourcekeyword match
Cole Murray/background-agents · TypeScript
Open-Inspect is an open-source background coding agent system inspired by Ramp's Inspect. It runs tasks in full sandboxed development environments with Node.js, Python, git, browser automation and VS Code. Sessions start from a web UI, Slack, GitHub PRs, Linear issues or webhooks, and support multiplayer collaboration, PRs attributed to the prompting user, scheduled and event-driven automations, and parallel sub-tasks in separate sandboxes. Models include Claude, OpenAI Codex, xAI Grok and OpenCode Zen. It is designed for single-tenant deployment only.
Open sourcekeyword match
Moltis/moltis · Rust
Moltis is a self-hosted personal agent server written in Rust and distributed as one binary that can run on hardware such as a Mac Mini or Raspberry Pi. Every command runs in a sandboxed container (Docker/Podman, Apple Container or WASM), and keys stay on the user's machine. Features include voice, memory with cross-session recall, edit checkpoints, scheduling, browser automation, MCP tools over stdio and HTTP/SSE, and channels including Telegram, Signal, Discord and Teams. Its README positions it against OpenClaw and Hermes Agent.
Open sourcekeyword match
Infisical/agent-vault · Go
Agent Vault is an open-source credential broker from Infisical that sits between agents and the APIs they call. Agents route HTTPS traffic through it via HTTPS_PROXY. The vault terminates TLS with a locally trusted CA, strips any credentials the agent sent, injects the stored credential, and opens a verified connection upstream. This covers APIs, CLIs, SDKs and MCP tools alike, with egress filtering and request logging. Infisical announced it as a research preview on April 22, 2026, for Claude Code, OpenClaw, Hermes and custom harnesses, and recommends its commercial Infisical Agent Vault for production.
Open sourcekeyword match
deeplethe/sandstorm · Rust
Sandstorm runs AI agent sandboxes on gVisor and tiers idle ones down (hot, frozen via cgroup freeze and zswap, snapshot to disk, detached) so a machine holds far more sandboxes than it could keep running; any request wakes a sandbox back to hot. A single sandstormd binary acts as master, node and web panel, with the master keeping desired state in SQLite. The README reports six 2-vCPU/2 GB nodes holding 20,000 sandboxes, and frozen wake times of roughly 100 ms.
Open sourcekeyword match
Kocoro Lab/Shannon · Go
Shannon is a multi-agent orchestration framework aimed at production use. It runs agents on Temporal workflows with time-travel replay debugging, enforces hard per-task and per-agent token budgets with model fallback, and supports human approval workflows and swarm collaboration. Code execution is sandboxed with WASI, policies use OPA, and observability comes via event streaming, Prometheus metrics and OpenTelemetry. It works with OpenAI, Anthropic, Google, DeepSeek, xAI and local models via Ollama, and exposes a REST API, Python SDK and desktop app.
Open sourcekeyword match
BitMiracle AI/Dormice · TypeScript
Dormice is a self-hosted sandbox platform for AI agents that treats sandboxes as permanent rather than disposable. An idempotent acquireSandbox(userKey) call creates, wakes, restarts or restores the same sandbox for a key, while idle sandboxes cool down through active, frozen, stopped and archived states. It deploys as one daemon with a SQLite ledger, uses Docker plus gVisor for execution and S3 for archives, and exposes an E2B-compatible API so the official e2b SDK works by changing two URLs. The project describes itself as in early development.
Open sourcekeyword match
MemovAI/mimiclaw · C
MimiClaw runs a personal AI assistant directly on an ESP32-S3 microcontroller in pure C, with no Linux or Node.js. Users message it over Telegram; the board receives the message via WiFi, runs an agent loop in which the LLM calls tools and reads local memory, and replies. It supports Anthropic and OpenAI as runtime-switchable providers, keeps memory across reboots and draws about 0.5 W from USB power. It is an example of a minimal embedded agent harness.
Open sourcekeyword match
elizaOS/eliza · TypeScript
elizaOS is an open-source TypeScript framework and product stack for autonomous AI agents. Its monorepo holds the core runtime, the Eliza app, a CLI, cloud services, native bridges and first-party plugins, with Linux and Android distribution tooling for running a whole device as elizaOS. Developers build agents and plugins against the runtime, can run local inference, and use a bundled benchmark orchestrator; a hosted Eliza Cloud is also offered.
Open sourcekeyword match
opensandbox-group/OpenSandbox · Python
OpenSandbox gives AI applications isolated environments to execute code, run commands, manage files and operate browsers or desktops on your own infrastructure. It starts locally with Docker and scales to Kubernetes through one sandbox lifecycle API, with a Firecracker-based fast runtime (reported 64 ms P50 creation) that supports pause/resume. Per-sandbox egress policies, an ingress gateway and a credential vault keep secrets out of workloads; SDKs cover Python, Java/Kotlin, TypeScript, C# and Go, plus a CLI and MCP.
Open sourcekeyword match
Tencent Cloud/CubeSandbox · Go
Cube Sandbox is a secure sandbox service from Tencent Cloud built on RustVMM and KVM. It is compatible with the E2B SDK and creates hardware-isolated sandboxes in under 60 ms, deploying on a single node or scaling to multi-node clusters, including Kubernetes. Releases have added the CubeCoW copy-on-write snapshot engine for cloning and rollback, a credential vault that keeps keys out of the sandbox, AutoPause/AutoResume, network policy hardening and cross-node pause/resume with an S3 backend.
Open sourcekeyword match
NVIDIA/OpenShell · Rust
OpenShell is NVIDIA's runtime for running autonomous agents with access to files, packages, APIs and credentials without unrestricted access to data, secrets or network. Each agent runs in an isolated sandbox where kernel-level controls confine file access and system calls and every network connection passes a policy check; agents never see real credentials. Policy changes are checked with formal verification before approval. It includes a gateway control plane, Kubernetes deployment via Helm, SDKs and agent skills.
Open sourcekeyword match
Astro (withastro)/flue · TypeScript
Flue is a TypeScript framework for building autonomous agents in the style of Claude Code and Codex rather than scripted LLM calls. Its built-in harness gives any model sessions, tools, skills (SKILL.md files), instructions, filesystem access and a sandbox — virtual, local or remote container — composed with hooks such as useModel, useSandbox, useSkill and useTool. Agents can run locally via CLI or be deployed. The repository lives under the withastro GitHub organization.
Open sourcekeyword match
superradcompany/microsandbox · Rust
Microsandbox runs untrusted workloads — AI agents, user code, plugins, CI jobs and scrapers — inside fast local microVMs with hardware-level isolation. It runs standard OCI images, offers Docker-like workflows, reports average boot times under 100 ms, can fork and snapshot live sandboxes and is embeddable in code without a setup server or long-running daemon. Secrets are kept out of the VM, and agents can create their own sandboxes via its agent skills and MCP server. Runs on Linux, macOS and Windows.
Open sourcekeyword match
Matt Pocock/sandcastle · TypeScript
Sandcastle is a TypeScript library (published as @ai-hero/sandcastle) for orchestrating AI coding agents in isolated sandboxes with a single sandcastle.run() call. It sandboxes the agent with a configurable branch strategy and merges the resulting commits back. Native sandboxes use the OS process sandbox (Bubblewrap on Linux and WSL2, Seatbelt on macOS) and are the documented default; it also ships Docker and Podman providers and cloud providers for Daytona, E2B, Fly.io Sprites, Modal and Vercel, plus APIs for custom ones. Agent providers cover Claude Code, Codex, Pi, Copilot, Cursor and OpenCode.
Open sourcekeyword match
agent-infra/sandbox · Python
AIO Sandbox packages a browser (with VNC), terminal, file system, VS Code Server, Jupyter and MCP into a single Docker container for AI agents, exposing them through one API on port 8080. It supports API-key authentication across its services, publishes pinned release images (including a mainland-China mirror), and documents cloud deployment behind a reverse proxy or ingress. The project links an accompanying arXiv paper and ships examples and an evaluation suite.
Open sourcekeyword match
Docker/docker-agent · Go
Docker Agent is an open-source agent builder and runtime from Docker Engineering, shipped as a docker CLI plugin (`docker agent run`). Agents are defined declaratively in YAML or HCL with a model, instructions, toolsets and sub-agents, and the runtime handles the LLM loop, tool calls and streaming. It supports built-in tools plus any MCP server, multi-agent delegation, RAG, and providers including OpenAI, Anthropic, Gemini, Bedrock and Docker Model Runner. Agents can be pushed to and pulled from OCI registries, and run as a TUI, headless CLI, HTTP/MCP server or over A2A.
Open sourcekeyword match
qhkm/zeptoclaw · Rust
ZeptoClaw is a small, security-focused personal AI assistant runtime packaged as one Rust binary (around 6MB) for running agents locally, at the edge or on a VPS. It bundles tools, long-term memory, channels (Telegram, Slack, Discord, WhatsApp and more), 18 LLM providers with retry and fallback, agent swarms that delegate to sub-agents, a plugin system, cost tracking and token budgets. Its README positions it against OpenClaw-ecosystem security incidents and highlights container isolation across six runtimes, a tool approval gate and prompt-injection detection.
Open sourcekeyword match
NullClaw/nullclaw · Zig
NullClaw is an autonomous AI assistant runtime written in Zig and compiled to a tiny static binary (678 KB, ~1 MB RAM, millisecond startup per its README) aimed at cheap ARM, x86 and RISC-V boards. It claims 50+ providers, 19 channels, 35+ tools, multiple memory engines, MCP, subagents and voice, with core systems exposed as swappable vtable interfaces. Security features include pairing, sandboxing via landlock, firejail, bubblewrap or Docker, allowlists, workspace scoping and encrypted secrets. It sits within a wider Null ecosystem (NullHub UI, nullboiler orchestration, nullwatch observability).
Open sourcekeyword match
Cloudflare/moltworker · TypeScript
MoltWorker is an experimental open-source project from Cloudflare that runs OpenClaw (formerly Moltbot/Clawdbot) inside a Cloudflare Sandbox container, fronted by an entrypoint Worker that routes and proxies requests. It uses the Sandbox SDK for isolated execution, Browser Rendering for headless Chromium via a CDP proxy, R2 for persistent storage, AI Gateway for model routing and Zero Trust Access for authentication. Cloudflare's launch post (Jan 2026) calls it a proof of concept rather than a Cloudflare product; it requires the Workers Paid plan.
Open sourcekeyword match
Rivet/agents · TypeScript
Rivet's agents repository ships two pieces of harness infrastructure. Rivet Agents runs agents such as Pi as durable Rivet Actors whose memory survives restarts. Sandbox Agent is a static Rust server that runs inside a sandbox (E2B, Daytona, Modal, Cloudflare Containers, Docker and others) and exposes one HTTP/SSE API to control Claude Code, Codex, OpenCode, Cursor, Amp or Pi, handling streaming events, permissions and sessions. It normalizes agent events into a universal session schema that can be persisted to Postgres, ClickHouse or Rivet for replay and audit, and also runs as an embedded TypeScript SDK.
Open sourcekeyword match
Coder/xum · TypeScript
Xum, renamed from Mux after a trademark concern, is an AGPL-3.0 desktop and browser app from Coder Technologies for running multiple coding agents in parallel, each in an isolated workspace with a central view of git divergence. Workspaces can run locally, in git worktrees, or on remote servers over SSH. It is multi-model and provider-agnostic, supports Ollama for local LLMs, and offers keybinds for managing many agents plus rich markdown output including Mermaid and LaTeX.
Open sourcekeyword match
Omnara/omnara · Go
Omnara (YC S25) is an Apache-2.0, self-hostable API for running agents in production, positioned as an alternative to Claude Managed Agents and OpenAI's Agents API. State lives in an event log so agents are durable, recover from failures and idle without compute. Agents run on Omnara machines, the user's Blaxel, Daytona or Unikraft accounts, or an outbound-connecting daemon. It adds human approval for risky tool calls, scoped secrets, RBAC and append-only audit history, exposes REST, TypeScript SDK and CLI, and works with OpenAI, Anthropic, OpenRouter, LiteLLM or self-hosted models.
Open sourcekeyword match
TabbyML/tabby · Rust
Tabby is an open-source, self-hosted AI coding assistant positioned as an on-premises alternative to GitHub Copilot. The TabbyML team also builds Pochi, a linked agent project that turns GitHub issues into tasks and creates PRs.
Open sourcekeyword match
dagger/container-use · Go
Container Use from Dagger provides containerized development environments for coding agents, so that multiple agents can work safely and independently with your preferred stack. It is Apache-2.0 licensed.
Open sourcekeyword match
anthropics/claude-code-action · TypeScript
A general-purpose GitHub Action that runs Claude Code on pull requests and issues. It can answer questions, review PRs and implement code changes. It picks an execution mode from the workflow context and runs on your own GitHub runner. It supports structured JSON outputs and the Bedrock, Vertex AI and Foundry providers.
Open sourcekeyword match
agent0ai/agent-zero · Python
Agent Zero is an open agent framework that gives the agent a Dockerized Linux desktop, a browser with DOM annotation, live document cowork, projects, skills and plugins. It supports per-project memory, multi-agent delegation to subagents, a plugin hub, and a CLI bridge to the host machine for working in local repositories. It runs wherever Docker runs.
Open sourcekeyword match
kortix-ai/suna · TypeScript
Suna is Kortix's open-source agent platform, described by the repo as 'The open-source AI Operating System'. It lets you self-host a general-purpose agent, and the project's homepage is kortix.com.
Open sourcekeyword match
amantus-ai/vibetunnel · TypeScript
VibeTunnel lets you use a web browser as a terminal, so you can control terminal-based coding agents remotely. Its homepage is vt.sh.
Open sourcekeyword match
Fosowl/agenticSeek · Python
The repo describes itself as a fully local Manus AI with no APIs or monthly bills. It is an autonomous agent that thinks, browses the web and writes code, costing only electricity to run.
Open sourcekeyword match
julep-ai/julep · Python
Julep provides durable, composable AI agent flows that can crash and resume, retry safely, and explain every step. It is Apache-2.0 licensed and written mainly in Python.
Open sourcekeyword match
iii-hq/iii · Rust
iii (by Motia LLC, formerly the Motia project) turns backend concerns such as queues, cron, HTTP, state, observability, agents and sandboxes into interoperable Workers, Triggers and Functions on one live runtime. Agents can add workers, discover their functions, call them and trace results. Its SPDX file declares Apache-2.0 AND Elastic-2.0.
Open sourcekeyword match
ogx-ai/ogx · Python
Drop-in replacement for the OpenAI API that runs on a laptop, in a datacenter or in the cloud, with a pluggable provider architecture (Ollama, vLLM, managed services). Its Responses API handles agentic orchestration on the server side, with tool calling, MCP server integration and built-in file search (RAG). Formerly Llama Stack.
Open sourcekeyword match
vndee/llm-sandbox · Python
Lightweight and portable sandbox runtime and code interpreter, as a Python library, for safely running LLM-generated code.
Open sourcekeyword match
aws/bedrock-agentcore-sdk-python · Python
Python SDK for turning any AI agent into a production-ready application. Gives framework-agnostic primitives for runtime, memory, authentication and tools, running on AWS-managed infrastructure (Amazon Bedrock AgentCore).
Open sourcekeyword match
docker/mcp-gateway · Go
Docker's MCP CLI plugin and MCP Gateway.
Open sourcekeyword match
nanobrowser/nanobrowser · TypeScript
Nanobrowser is an open-source AI agent for web automation. It runs in your own logged-in browser to automate daily tasks.
Open sourcekeyword match
OthersideAI/self-operating-computer · Python
Self-Operating Computer is a framework from OthersideAI that lets a multimodal model operate a computer. The model reads the screen and controls the keyboard and mouse.
Open sourcekeyword match
microsoft/magentic-ui · Python
MagenticLite, formerly Magentic-UI, is an experimental agent from Microsoft that works across the browser and the local file system. The repo's homepage links to a Microsoft Research blog post on MagenticLite, MagenticBrain and Fara1.5, which describes an agentic experience optimized for small models.
Open sourcekeyword match
browseros-ai/BrowserOS · TypeScript
BrowserOS is an open-source agentic browser, positioned as an alternative to ChatGPT Atlas, Perplexity Comet and Dia.
Open sourcekeyword match
lightpanda-io/browser · Zig
Lightpanda is a headless browser designed for AI and automation. The repo's primary language is Zig.
Open sourcekeyword match
superhq-ai/superhq · Rust
A Rust/GPUI desktop app for macOS that runs multiple AI coding agents (Claude Code, OpenAI Codex, Pi) in isolated sandbox VMs with full terminal access. Agents never see the real API keys: an auth gateway on the host injects credentials into outgoing requests. The README describes it as very early alpha.
Open sourcekeyword match
dtormoen/tsk-tsk · Rust
Delegates development tasks to YOLO-mode Claude Code or Codex agents running in parallel in filesystem- and network-isolated Docker or Podman containers. It auto-detects the toolchain, builds container images, copies the repo without gitignored files, and enforces a domain allowlist through a proxy sidecar. Agent branches are fetched back to the repo for review.
Open sourcekeyword match
JuliusBrussee/caveman · Go
Caveman makes coding agents answer tersely and read less, which it says cuts about 65% of tokens. One npx install sets up the skill and a local proxy, so plain claude, Codex, Gemini CLI, OpenCode and 30+ other agents stay caveman'd. Each of its six parts has its own switch, and the skill can be installed without the proxy.
Open sourcekeyword match
numman-ali/opencode-openai-codex-auth · TypeScript
OpenCode plugin that signs in with OpenAI's official OAuth flow, so a ChatGPT Plus/Pro subscription can be used for coding.
Open sourcekeyword match
Mintplex-Labs/anything-llm · JavaScript
Local-first application for running a private agent setup, including RAG, multimodal input, computer use, and agent orchestration. It is self-hosted rather than a rented hosted service.
Open sourcekeyword match
CherryHQ/cherry-studio · TypeScript
Open-source AI workspace for models, agents, and productivity. Topics include an MCP client, agent skills, computer use, RAG, and Ollama.
Open sourcekeyword match
yc-software/qm · TypeScript
qm is a TypeScript project described as a multiplayer agent harness for teams. It is MIT-licensed, and its homepage is on Y Combinator labs. Public topics are ai, assistant, harness, and qm.
Open sourcekeyword match
NanmiCoder/cc-haha · TypeScript
cc-haha is a local-first, cross-platform desktop workspace for Claude Code and AI agents. It includes multi-agent collaboration, Git worktrees, code diffs, a skills marketplace, multi-model support, and Computer Use. Remote access is described via WeChat, Feishu, DingTalk, Telegram, WhatsApp, and H5.
Open sourcekeyword match
TencentCloud/Octop · Python
Octop is a self-hosted AI assistant for multi-user, multi-agent use. Repository topics include agentic AI, local-first deployment, and long-term memory.
Open sourcekeyword match
alibaba/spring-ai-alibaba · Java
Spring AI Alibaba is an agentic AI framework for Java developers. Topics cover multi-agent workflows, graphs, ReAct-style agents, and Spring AI.
Open sourcekeyword match
HarnessMD/munder-difflin · JavaScript
Open-source app for running an office of Claude Code, Codex, and similar agents on a laptop, in sandboxes, or elsewhere, using existing subscriptions. It targets local-first multi-agent orchestration, memory, and coding-agent workflows.
Open sourcekeyword match
Orkas-AI/Orkas · JavaScript
Orkas is an MIT-licensed, local-first desktop AI team for macOS and Windows. A Commander turns a goal into to-dos and coordinates 16 built-in specialist agents plus Claude Code, Codex, OpenCode, OpenClaw, and Hermes, with projects, scheduled automations, and connectors. It can use built-in models or your own keys.
Open sourcekeyword match
iflytek/astron-agent · Java
Open-source agentic workflow platform from iFlytek for multi-agent orchestration, MCP, and workflow automation. It includes low-code and workflow-engine components and is released under Apache-2.0.
Open sourcekeyword match
YaoApp/yao · Go
Self-hosted app for keeping agents and workspaces in one place and tracking tasks on a board. It is accessible from desktop, mobile, browser, or API, and is positioned around agent harnesses, orchestration, skills, and coding-agent workflows.
Open sourcekeyword match
vastsa/PI-Desktop · TypeScript
PI-Desktop is a local-first AI coding agent desktop. It uses Electron, a Rust host core, and the pi Agent Harness, and supports user-installable plugins and MCP.
Open sourcekeyword match
loopx-project/loopx · Python
Control plane for always-on agents that turns long-horizon goals into verifiable results across Codex, Claude Code, and custom runtimes. Ranges from local SQLite to team scale. Python, Apache-2.0.
Open sourcekeyword match
agentscope-ai/agentscope-java · Java
AgentScope Java is an open-source framework for building distributed, production-grade, long-running agents. It is written in Java and licensed under Apache-2.0.
Open sourcekeyword match
KunAgent/Kun · TypeScript
Kun is a local-first AI agent workspace for coding, writing, design, research, and automation. It uses one runtime for a desktop GUI and a TUI, and its topics include MCP, multi-agent workflows, and coding agents. It is written in TypeScript.
Open sourcekeyword match
microsoft/CUAWright · Python
CUAWright is a Python framework for browser and desktop agents, described as a simple SWE-style framework for long-horizon computer-use tasks. It is published by Microsoft under the MIT license.
Open sourcekeyword match
cloudflare/vibesdk · TypeScript
An open-source vibe-coding platform for building your own vibe-coding platform, built entirely on the Cloudflare stack. Repository topics include coding-agent, text-to-app, Cloudflare Workers, and Durable Objects.
Open sourcekeyword match
ApodexAI/FrontierAgent · Python
Open-source agent framework with a native command-line TUI, ReAct and Agent Team modes. It is started with one command on macOS and Linux, with no preinstall and no hard Docker dependency.
Open sourcekeyword match
sentient-agi/ROMA · Python
Recursive-Open-Meta-Agent v0.1 (beta), a meta-agent framework for building multi-agent systems. Published by Sentient AGI.
Open sourcekeyword match
SolaceProducts/solace-agent-mesh · Shell
Platform for building and running AI agents across build, test, deploy, monitor, and improve. It hosts agents, runs their tools, and coordinates them over an event mesh, with sessions, artifacts, auth, evaluators, and observability built in.
Open sourcekeyword match
TencentCloudADP/youtu-agent · Python
A Python agent framework for building agents that deliver with open-source models. Repository topics also include OpenAI-style agents.
Open sourcekeyword match
embabel/embabel-agent · Kotlin
Agent framework for the JVM, implemented in Kotlin, for multi-agent systems and orchestration. Apache-2.0 licensed, with Spring among its stated stack.
Open sourcekeyword match
CommandCodeAI/command-code · Unknown
Command-line coding agent whose repository topics describe a harness for models including DeepSeek, GLM, Kimi, OpenAI, and Anthropic. The public description does not list further features.
Open sourcekeyword match
NVlabs/SoL-Pi · TypeScript
NVlabs project on scaling auto-research loops for efficient agent harnesses. TypeScript codebase under the MIT license, with a project page.
Open sourcekeyword match
DSH-APP/DSHA · Java
Android project for running DeepSeek Harness on a phone without root or Termux. It describes a full Ubuntu environment with proroot, live on-screen AI output, direct ADB access, and retained data.
Open sourcekeyword match
fuxicodex/Fuxi · Python
FuXi is a self-contained terminal AI coding agent for editing code, running commands, and driving tools. It includes cost-aware routing across LLM providers and lists MCP among its topics.
Open sourcekeyword match
MiroMindAI/MiroFlow · Python
MiroFlow is a Python agent framework that includes a web UI and supports MiroThinker, Claude, Kimi, and OpenAI. Repository topics center on research agents and agent benchmarks.
Open sourcekeyword match
SciSharp/BotSharp · C#
BotSharp is an AI multi-agent framework in .NET, written in C#. Its listed focus is chatbots and multi-agent systems.
Open sourcekeyword match
Pinvou/pinvou-agent · Rust
pinvou-agent is an open-source desktop AI agent for tools, files, knowledge, workflows, and deliverables. Topics include a Rust app, MCP client support, local-first use, and Ollama.
Open sourcekeyword match
itayinbarr/little-coder · TypeScript
A harness optimized for smaller LLMs and coding agents. Topics cover local models, Ollama, tool use, and Terminal-Bench.
Open sourcekeyword match
citrolabs/ego-lite · JavaScript
A browser for AI-agent browser automation that can share a logged-in browser session with coding agents such as Codex or Claude Code. It is published as an agent skill for those harnesses.
Open sourcekeyword match
agentuniverse-ai/agentUniverse · Python
agentUniverse is an Apache-2.0 Python framework for building LLM multi-agent applications. The repo is tagged for agents, autonomous systems, and multi-agent workflows.
Open sourcekeyword match
espressif/esp-claw · C
ESP-Claw is an Apache-2.0 C project from Espressif described as a chat-coding AI agent framework for IoT devices. A project site is published at esp-claw.com.
Open sourcekeyword match
OpenCoworkAI/open-cowork · TypeScript
open-cowork is an MIT-licensed desktop app for Windows and macOS that one-click installs Claude Code, MCP tools, and skills. It includes sandbox isolation, multi-model support, and Feishu and Slack integration.
Open sourcekeyword match
skalesapp/skales · Unknown
Skales is a personal AI agent harness for macOS, Windows, Linux, Android, and iOS, with chat, coding (Skales Code), Flow, automations, computer use, browser, voice, MCP, BYOK across 26 providers, and local LLMs via Ollama. Skales Scout runs lasting desk agents, optionally each in its own virtual machine. It does not use Docker or a terminal.
Open sourcekeyword match
unreallabsai/unreal-agent · Go
unreal-agent is an async-first agent harness written in Go and released under the MIT license.
Open sourcekeyword match
ShenSeanChen/waku-agent · Python
Waku Agent is a local-first AI agent harness that includes the loop, memory, and eval in code intended to stay legible as it grows. It is written in Python and released under the MIT license.
Open sourcekeyword match
AgnesAI-Labs/agnes-harness · TypeScript
Agnes Harness is a pluggable TypeScript agent harness for forward deployed engineering (FDE). It is released under the Apache-2.0 license.
Open sourcekeyword match
zhnt/loushang · Python
Loushang is a Python agent harness for coding workflows. It covers multi-model LLM orchestration, stateful sessions, tool governance, traceable delivery, and provider routing for GPT, Claude, DeepSeek, Qwen, Kimi, GLM, and MiniMax.
Open sourcekeyword match
yomorun/yomo · Rust
A serverless AI agent framework with geo-distributed edge AI infrastructure. Repository topics include MCP, A2A, skills, function calling, Claude Code, and QUIC.
Open sourcekeyword match
agentjido/jido · Elixir
An autonomous agent framework for Elixir, built for distributed autonomous behavior and dynamic workflows.
Open sourcekeyword match
Roy3838/Observer · TypeScript
Observer is an open-source TypeScript micro-agent orchestrator that runs completely locally and monitors things for you. It is licensed under AGPL-3.0.
Open sourcekeyword match
bubbuild/bub · Python
bub is a tiny Python agent runtime that is composable with plugins and can be self-hosted. It is licensed under Apache-2.0 and described as a harness.
Open sourcekeyword match
exoharness/exo · Rust
Exo is an agent and harness architecture that is fully recursive and can edit all aspects of itself at runtime to get better at tasks. It is written in Rust and licensed under MIT.
Open sourcekeyword match
AgentEra/Agently · Python
Agently is a Python framework for building GenAI applications and interacting with agents in code through structured data and chained calls. It uses an event-driven flow, TriggerFlow, for working logic and can switch models without rewriting application code. It is licensed under Apache-2.0.
Open sourcekeyword match
langchain-ai/deepagentsjs · TypeScript
TypeScript project from langchain-ai described as a batteries-included agent harness. It is MIT-licensed and documented under the LangChain JavaScript Deep Agents overview.
Open sourcekeyword match
agentlas-ai/Agentlas-OS · Python
Python agent OS that keeps specialist agents in a hub and spins up a temporary orchestrator per task. It is local-first, works with any model, and lists MCP and multi-agent topics. Apache-2.0.
Open sourcekeyword match
ThousandBirdsInc/chidori · Rust
Chidori is a Rust agent framework where runs are durable, replayable, and resumable by default. It includes checkpointing, deterministic execution, and replay, plus a JavaScript engine with TypeScript support.
Open sourcekeyword match
thClaws/thClaws · Rust
thClaws is an open-source AI agent harness written in Rust. One binary provides GUI, CLI, headless, and webapp interfaces, with multi-provider support, MCP, skills, plugins, and agent teams.
Open sourcekeyword match
rasbt/mini-coding-agent · Python
A minimal, readable Python implementation of a coding agent harness, written to explain the core components of coding agents. Licensed Apache-2.0.
Open sourcekeyword match
Azure/agent-landing-zone · Python
Landing zone for AI agent applications on Azure. It can deploy the infrastructure, the agent app, or both, using Microsoft Foundry and Microsoft Agent Framework, with Zero-Trust networking, observability, and infrastructure as code. MIT licensed.
Open sourcekeyword match
polyuiislab/infiAgent · Python
A Python agent framework for long-running, complex tasks. Agents are set up by editing config files, and it supports Anthropic skills. Licensed GPL-3.0.
Open sourcekeyword match
JetBrains/thinkrail · TypeScript
ThinkRail is a JetBrains project described as an agentic IDE and a pi-agent extension. The TypeScript repo is Apache-2.0 and tagged for agent workflows and spec-driven development.
Open sourcekeyword match
AMAP-ML/LongHorizon-Harness · Python
A computer-use harness for running AI agents across desktop apps and the CLI for extended periods while preserving task state. It provides fresh-context execution, durable verified state, independent auditing, and recoverable progress, with native Claude Code, Codex, and OpenClaw integration.
Open sourcekeyword match
hellowind777/hello2cc · JavaScript
A native-first Claude Code plugin for third-party models, with silent Agent model injection and output styles. JavaScript, Apache-2.0.
Open sourcekeyword match
shekohex/opencode-pty · TypeScript
An OpenCode plugin for interactive PTY management. It runs background processes, sends input, and reads output with regex filtering.
Open sourcekeyword match
SponsioLabs/Sponsio · Python
Open-source project described as deterministic safety solutions for probabilistic AI agents. Topics include guardrails, a policy engine, intent verification, runtime safety, and prompt injection. Packaged as a skill for OpenClaw.
Open sourcekeyword match
prekuter/dryforge · Python
Dryforge is a bounded-autonomy plugin harness for agents, described as taking work from intent to implementation. It is a Python plugin for Claude Code, Codex, GitHub Copilot, and Antigravity.
Open sourcekeyword match
aerovato/container · TypeScript
Persistent Linux workspaces for agentic development, as a plugin for Claude Code, Codex, OpenCode, and Copilot.
Open sourcekeyword match
eclipse-theia/theia · TypeScript
Eclipse Theia is an open-source platform for building custom IDEs and tools for browser and desktop. It includes Theia AI, the vendor-neutral AI harness platform, and powers the AI-native Theia IDE.
Open sourcekeyword match
mdiri-ai/SoduWork · JavaScript
SoduWork is an open-source, self-hosted engine for teams and their agents. It is described as multi-harness and multi-agent, with topics covering workflows, MCP, a plugin system, human-in-the-loop collaboration, and harnesses such as Claude Code, Codex, and OpenCode.
Gatheringkeyword match
Feb 3–4, 2027 · San Francisco, CA
Daytona's conference at Chase Center on the infrastructure layer powering AI, agents and the next generation of cloud computing. Times on the event page may be placeholders.
Gatheringkeyword match
Nov 9–10, 2026 · Online
A free two-day virtual conference on safety, security and governance for autonomous agents: sandboxing, prompt-injection defence and running agents at scale, with 5 hands-on tutorials.