Open-source platform for multi-agent workflows

An open-source platform for multi-agent workflows has been launched, allowing each agent to run in its own isolated container on the user's cloud. The platform supports parallel execution of 5 to 500 agents, real-time visibility into agent actions, and persistent project memory, enabling users to manage and monitor agent tasks from a single interface.

The open-source platform for multi-agent workflows Every agent runs in its own isolated container on your cloud. Run 5 or 500 in parallel — SAM handles provisioning, scheduling, and cleanup. Describe what you want. Watch it happen. Talk to your agents like teammates. Pick the right model for the job, attach files, and watch them write code, run tests, and open PRs — all from a single chat. Multiple agents, one interface Switch between Claude Code, Codex, Gemini CLI, Amp, and more — each configured with their own API key or OAuth token. Choose the right agent for the task without leaving the chat. Explore Chat-First Development /features/chat/ Real-time visibility into agent work Tool calls, file edits, and shell commands are streamed to your browser as they happen. You can read every step the agent takes and cancel at any time. Explore Chat-First Development /features/chat/ See everything. Stay in control. A dashboard for all your agent work. Track active tasks across projects, monitor node health, and get notified when agents need your attention. Multi-project dashboard See active tasks with their status, project, and timing. Below that, a project grid shows workspace counts, session counts, and last activity for each repo. Explore Full Visibility & Control /features/visibility/ Infrastructure you can see Each node card shows its provider, spec, pricing, CPU/memory/disk usage, and the workspaces running on it. Create new nodes or add workspaces from the same page. Explore Full Visibility & Control /features/visibility/ Never miss a decision point Notifications are categorized by type — task completions, errors, progress updates, sessions ended, PRs created, and moments where agents need human input. Filter to focus on what matters. Explore Full Visibility & Control /features/visibility/ Agents that remember your preferences. Build a project knowledge base that agents receive as context. Set policies that are injected into every session. Review a full log of agent activity. Persistent project memory Store observations about your architecture, conventions, and preferences. Each entry has a confidence score, a source explicit or inferred , and a confirmation date so you can see what the knowledge base contains. Explore Agents That Learn /features/agent-context/ Project policies Define project-level instructions like "use portals for modals" or "CLI packages must meet QA standards." Policies are tagged as preferences or rules and injected into every agent session as context. Explore Agents That Learn /features/agent-context/ Activity log A timestamped feed of every task submission, execution step, session start/stop, and agent completion event. Filter to find specific activity across the project. Explore Agents That Learn /features/agent-context/ From idea to agent-ready task. Capture ideas with a problem statement and plan. Schedule recurring tasks on a weekly or daily cadence. When you are ready, dispatch an agent with one click. Structured idea capture Write down the problem, sketch a plan, and link related chat conversations for context. When you are ready, hit Execute to dispatch an agent that picks up the idea as a task. Explore Idea to Execution /features/automation/ Recurring triggers Schedule agents to run dependency updates, spot checks, audits, or other recurring work on a daily or weekly cadence. Pause, resume, or run manually any time. Explore Idea to Execution /features/automation/ Fully configurable. Open source. Upload reference docs, choose your VM size, and pick a default agent — each project has its own settings for infrastructure and agent configuration. Project library Upload architecture docs, research, and reference material. Files are organized with folders and tags. Agents can access library files as additional context during their work. Explore Your Project, Your Way /features/configuration/ Built-in document viewer Markdown files render with full formatting, including Mermaid diagrams. Toggle between the rendered view and raw source, or download the file directly. Explore Your Project, Your Way /features/configuration/ Infrastructure and agent defaults Choose a default node size Small, Medium, or Large with specs and pricing shown and a default agent type for each project. Per-agent credential overrides let you configure API keys at the project level. Explore Your Project, Your Way /features/configuration/ From zero to coding agents in minutes Four steps to coding agents working through your backlog on cloud VMs. Connect Your Cloud Add your Hetzner, Scaleway, or GCP credentials. Encrypted per-user. Create a Project Link a GitHub repo. SAM organizes tasks and workspaces per project. Describe Your Task Type what you need. SAM provisions a VM, builds your devcontainer, and runs the agent. Watch It Build Stream agent output in real time. Get notified when the agent needs input or finishes. How SAM stacks up Own your infrastructure. Choose your agent. Pay cloud prices. | SAM | Managed AI Platforms Devin, Ona | Self-Hosted CDEs Coder | | |---|---|---|---| | Your code runs on | Your cloud Hetzner, Scaleway, GCP | Their infrastructure | Your infra AWS, GCP, K8s | | Agent choice | Claude Code, Codex, Gemini CLI, Mistral Vibe, OpenCode, Amp | Their proprietary agent | Varies Claude, OpenAI | | Open source | AGPL-3.0 license | Proprietary | Open-source core | | Control plane | Free Cloudflare Workers | Bundled into pricing | You host it K8s / Docker | | Infrastructure cost | ~$4–15/mo direct to provider | From $20/mo compute + inference | Your cloud bill + license fees | | Self-hostable | Fully — control plane + VMs | No | Yes — Terraform + K8s required | | Setup effort | Deploy to Cloudflare + connect cloud | Zero config | Terraform + cluster management | | Data sovereignty | Code runs on your cloud; control plane data in Cloudflare | Vendor-managed | Full — your infrastructure | SAM infrastructure costs are compute only — bring your own API keys for AI providers. Managed platform pricing bundles compute and model inference. Built in the open, shipped fast Nine phases shipped. More providers and team collaboration coming next. Workspace management, GitHub OAuth, encrypted credentials, VM provisioning WebSocket terminal, structured agent chat via ACP Claude Code, Codex, Gemini CLI, Mistral Vibe, OpenCode, Amp Multiple isolated workspaces per VM Projects, persistent chat, activity feeds, task-driven workflows Scaleway, GCP, notifications, voice input, conversation forking, warm pooling Missions, agent-to-agent dispatch, task dependencies, durable messaging Auth, task submission, chat prompts, task status, and runner doctor commands Deployment environments, agent policy gates, Compose releases, logs, config, and volumes DigitalOcean, Vultr, AWS, and expanded provider coverage Team workspaces, shared projects, organization management Ready to run multi-agent workflows at scale? Self-host on low-cost Cloudflare Workers. Bring your own cloud. Your agents, your infrastructure.