Dash Launches Runtime Protection for Coder

Dash Launches Runtime Protection for Coder

BY Eran Hod

5 MIN READ

A developer laptop’s SSH session passes through Dash into a Coder workspace running Claude Code

AI agents don't stay in one place. A developer starts Claude Code on a laptop, continues in a browser, and hands the heavy work to an agent in a remote workspace. Security teams need one set of rules that follows the agent the whole way, and one record of who was behind it.

Today Dash extends runtime protection to Coder, the self-hosted platform many enterprises use to run development and AI agents on their own infrastructure. Coder decides where agents run and what they can reach. Dash decides what they are allowed to do, on every prompt, shell command execution and tool call, and ties each action to a real person and device.


Protection on every agent action

Dash checks what agents in Coder are about to do before they do it. For Coder Agents, Dash connects at the Coder server, so there is nothing to install in any workspace. Claude Code, Codex and other agents running inside workspaces are protected by the Dash Endpoint client.

  • Secrets never reach the model: A password or API key in a prompt is removed on the fly, and the agent keeps working without it.

  • Out-of-policy actions never run: A risky command or tool call is stopped before it executes, and the agent is told why.

  • Unapproved models never answer: If someone picks a model outside policy, they get a clear Dash notice in the chat instead of a response.

  • Hijacked agents get blocked: A README, a tool result or a web page can carry hidden instructions that hand the agent a new objective. Dash detects the injected prompt and blocks it before the agent acts on it.

Know who is behind every session

Inside a remote workspace, every agent looks the same: one shared system user, often a shared API key, and sometimes a whole team working in the same workspace. Dash links each session back to the developer and the laptop it came from, whether they connected from a terminal, an IDE, Coder Desktop or a browser.

Every session in Dash shows which workstation it was connected from, linked to that device in Endpoint Inventory, and policies apply to the person instead of the shared account.


One investigation, from laptop to workspace

When something goes wrong, Dash shows the whole story in one place: who started the session, from which device, what was asked, which tools ran and where policy stepped in. The same view already covers agents on workstations, in browsers and in SaaS tools, so a session that crosses all of them is still one investigation.

See and control Coder across your fleet

Dash finds every Coder CLI and Coder Desktop install on macOS, Windows and Linux, with versions, the Coder deployment each one connects to, and the person using it. Security teams can allow approved use and block the rest through the Dash Agent or EDR.

Simple to roll out

Connect Coder from the Dash console and follow the guided setup. Dash generates ready-to-use configuration for however you run Coder: Helm, Kubernetes, Docker or Terraform. Teams that manage Coder with Terraform add Dash to their existing configuration and can roll it out across every Coder deployment the same way.

Start with a pilot group, confirm what Dash sees, then expand to everyone. Developers keep working exactly as they do today.

FAQs

Which agents does Dash protect, and do we need to install anything?

Dash protects Coder Agents, plus the coding agents developers run inside workspaces, such as Claude Code, Codex and IDE agents. Coder Agents need nothing installed in the workspace, because Dash connects at the Coder server. Agents running inside workspaces are covered by the Dash Endpoint client on the workspace.

How does Dash know who is behind a shared workspace?

Dash matches each session to the laptop it was connected from, whether the developer came in through a terminal, an IDE, Coder Desktop or the browser, and to that person's corporate identity. Even when a team shares one workspace under a service account, sessions point to the person, and Dash shows how each match was made.

Does Dash work with Coder's AI Gateway and Agent Firewall?

Yes, and without them too. Dash covers the same ground on its own: model and MCP controls, limits on the destinations agents reach, and a per-user record of every prompt and tool call. On top of that, it inspects what agents send, decides on every action before it runs, and ties each session to a person and device.

Will developers notice Dash, and what if Dash is unavailable?

Developers keep working in the CLI, Coder Desktop, their IDE or the browser as before, and only hear from Dash when a policy steps in, with a clear reason why. Coder waits only briefly for each Dash check, so if Dash ever can't respond, the agent continues and work is never blocked.

Can we roll out gradually?

Yes. Start with a pilot group of users or teams, review what Dash sees, then widen the scope from the Dash console. Teams with several Coder deployments can manage all of them through Terraform.

See what your agents are doing.

© 2026 Dash Security, Inc. All rights reserved.

© 2026 Dash Security, Inc. All rights reserved.

© 2026 Dash Security, Inc. All rights reserved.

© 2026 Dash Security, Inc. All rights reserved.