Wirken Releases Open-Source AI Agent Gateway
The tool links nine messaging platforms to AI agents while limiting what a compromised agent can reach.
Updated on Nov. 9, 2026 in Artificial Intelligence
Wirken has released an open-source gateway that sits between messaging apps and AI agents, connecting Signal, WhatsApp, Telegram, Matrix, Discord, Slack, Teams, Google Chat and iMessage, plus a web chat. The software is available for free on GitHub.
Why it matters
Any AI agent can be turned against its owner, so the gateway is built to restrict what a compromised agent can reach. Each messaging channel runs as a separate process holding only its own credentials, and credentials are supplied at the edge so neither the agent nor the model handles them.
Shell commands run in an ephemeral container with no network access, and every action is written to a signed, hash-chained log before it runs. Sample audit output shows 1234 rows across 7 sessions with 24 verified chain-head signatures, though independent verification of these figures is not reported.
The players
Wirken
Wirken is the open-source project behind the messaging-to-agent gateway, distributed free through GitHub with a shell installer for Linux and macOS and a separate path for Windows 11.
GitHub
GitHub is the Microsoft-owned code hosting platform where Wirken is published and where developers can download and inspect its source code.
The details
Actions that change something ask for approval the first time and remember the answer, while deleting, installing or touching credentials asks every time. Log entries are attributed to the person who triggered the action and can be verified offline, and an altered entry fails verification. Shell execution is refused if the sandbox is unavailable, and Docker is required for the sandbox, which is on by default.
Timeline
Wirken's open-source release was published on GitHub on November 9, 2026.
The Tech Race
Wirken applies Docker container isolation, a mature tool from software development, to the newer problem of constraining AI agents that can execute commands. The gateway's design of separate per-channel processes and an append-only signed audit log positions it within a growing effort to keep agents from being turned against their owners.
Users who connect messaging apps to AI agents gain the ability to run shell commands and make changes, but the gateway asks for approval before anything that alters data or credentials. Setup requires Docker for the default sandbox, and users without it can still use other features but will lose shell command access.
The takeaway
Developers building or using AI agents can review Wirken's approach of separating credentials, sandboxing commands and logging actions as a practical model. Checking the installer's published hash before running it is advised.
Further reading
For broader context on how these tools are reshaping software, explore our Artificial Intelligence coverage.
More information
Download the free open-source project from Wirken on GitHub.
Source note: This article includes information reported by Help Net Security.
