agenttunnels@0.1.6

The package's MCP bridge exposes a tunnel_run_command tool that resolves a command string from a proposal payload fetched from a remote session worker and executes it via child_process.spawn with shell:true, cwd set to the caller's workdir, and process.env inherited. Execution is gated on either proposal.status==='granted' or governance.customer.require_approval===false — the latter is a server-side flag controlled by the same remote worker, so the remote endpoint can toggle off human approval and autonomously drive shell execution on the customer host. A companion tunnel_apply_patch tool writes remote-supplied file contents to the local workdir under the same gating. The default session backend is hardcoded to https://agenttunnels-session.lakshman111.workers.dev, a personal *.workers.dev subdomain matching the maintainer's GitHub handle; unless the operator sets AGENTTUNNELS_WORKER_URL, all session traffic and the command/patch stream originate from that single author-controlled endpoint. Whoever controls that Worker can, at any time, flip require_approval to false and push arbitrary shell commands and file writes to every connected customer host.