Ssh Exec

Run a single command on a remote Tailscale node via SSH without opening an interactive session.

Security Scan

The skill's stated purpose (run a single SSH command) matches its instructions, but the runtime instructions reference local SSH credentials and environment variables that the registry metadata does not declare and allow the agent to run arbitrary remote commands — this mismatch and the access to local SSH artifacts are concerning.

Purpose & Capability

The name/description (run a single command on a remote Tailscale node via SSH) aligns with the instructions: all examples are plain ssh invocations and piping scripts to ssh. The declared required binary (ssh) is appropriate for the stated purpose.

Instruction Scope

SKILL.md explicitly tells the agent to use local SSH keys (~/.ssh/) or the SSH agent socket (SSH_AUTH_SOCK) and to read environment variables like SSH_TARGET, SSH_HOST and SSH_PORT to determine the remote target. Those references broaden the skill's scope (access to local credentials and arbitrary remote command execution) and are not fully reflected in the metadata. The examples allow piping arbitrary scripts to remote hosts, which is functionally necessary for the skill but increases risk because the agent can execute arbitrary commands on reachable machines.

Install Mechanism

This is an instruction-only skill with no install spec and no code files, so nothing is written to disk by the skill itself — lower install risk.

Credentials

Registry metadata lists no required env vars or config paths, but SKILL.md references SSH_AUTH_SOCK, SSH_TARGET, SSH_HOST, and SSH_PORT. SSH_AUTH_SOCK and ~/.ssh are sensitive local credential artifacts; the skill implicitly expects access to them but does not declare them. That mismatch is disproportionate and should be clarified. The skill does not require a new API key, but it implicitly relies on the user's SSH credentials to perform its function.

Persistence & Privilege

always is false and there is no install that modifies system/agent config. However, because the platform allows autonomous invocation by default, an agent could use this skill to run SSH commands if permitted — combine this with the credential access noted above when deciding whether to allow autonomous use.

What to consider before installing

This skill is coherent for running single SSH commands, but before installing: 1) Confirm you trust the skill owner — the skill will (per its instructions) use your local SSH keys or SSH agent socket if available, which are sensitive. 2) Ask the owner/maintainer to update metadata to declare the env vars/config it uses (SSH_AUTH_SOCK, SSH_HOST/PORT/TARGET) so you can make an informed decision. 3) If you enable the skill, avoid exposing broad credentials: use a dedicated, limited SSH key and targets restricted server-side (authorized_keys forced-commands or limited user accounts). 4) Be cautious about allowing autonomous invocation — an agent could run arbitrary commands on remote hosts accessible via your SSH credentials. 5) If unsure, test in an isolated environment or refuse to provide SSH_AUTH_SOCK or private keys.

SSH Exec Skill

Run a single command on a remote Tailscale node via SSH without opening an interactive session. Requires SSH access to the target (key in ~/.ssh/ or SSH_AUTH_SOCK) and SSH_TARGET env var (e.g., 100.107.204.64:8022).

Execute a Remote Command

Run a command on the target and return stdout/stderr:

ssh -p 8022 user@100.107.204.64 "uname -a"

Execute with Custom Port

Use the SSH_TARGET env var:

ssh -p "${SSH_PORT:-22}" "$SSH_HOST" "df -h"

Run a Script Remotely

Pipe a local script to the remote host:

ssh -p 8022 user@100.107.204.64 'bash -s' < local-script.sh