dsh-remote connects out from your machine to SSH hosts you register, and lets an agent run shell commands there as your user. It deliberately does not expose the DSH Web UI on a public interface — that design is intentional (see the README's Safety section).
Practical implications you should be aware of:
- Adding a machine grants the agent arbitrary command execution as your user on that host over SSH. Only add machines you trust.
- Credentials (password or key passphrase) are stored locally in the machine file, or in the OS keychain when "加密保存密码" is enabled (macOS Keychain / Windows DPAPI / Linux
secret-tool). If no keychain backend is available, the plugin falls back to plaintext on disk — treat that file as sensitive and lock down its ACL. - When
auditLogis enabled, every executed command is recorded in the audit log; review it from the Settings page. - Remote files are mirrored into a local workspace. A compromised remote host can therefore influence what lands in your local mirror.
Only the latest published version on npm receives fixes. If you're on an older version, please upgrade before reporting.
Please do not open a public issue.
Report privately via GitHub's private vulnerability reporting on this repository. If that is unavailable to you, email flyphp@outlook.com.
Please include:
- affected version (
npm ls dsh-remote), - a description of the impact and the attacker model (who has to do what),
- reproduction steps or a proof of concept,
- any suggested fix.
Do not include real credentials, real private keys, or hostnames of machines you do not own. Redact them.
This is a spare-time project, so please allow a reasonable window. You'll get an acknowledgement, an assessment, and — for valid reports — a fix and credit in the release notes unless you prefer otherwise.
- Misconfiguration on your side (weak passwords, exposing your own SSH agent, adding machines you shouldn't).
- Vulnerabilities in
ssh2, Node.js, DeepSeek Harness, or the remote host's SSH daemon — please report those upstream. - Anything that requires an attacker to already have local code execution or your keychain unlocked.