Tencent Cloud CVM

腾讯云 CVM 云服务器运维工具集

This package is a full CLI/bash toolkit for Tencent Cloud CVM and will require your Tencent API keys and instance passwords to work — but the registry metadata did not declare those requirements. Before installing or running it: 1) Verify the source and trustworthiness of this bundle (no homepage provided). 2) Inspect the scripts locally (they are included) and confirm you are comfortable with plaintext password storage at ~/.tencent_cvm_passwords and the fact that scripts print passwords to stdout. 3) Prefer SSH key-based access over sshpass/passwords; if you must use passwords, restrict the password file (chmod 600) and consider storing secrets in a dedicated secret manager. 4) Be aware scripts can perform service management (systemctl) and file transfers — run them manually and avoid granting broad automation privileges. 5) If you expect the skill to be used by an agent, ensure the agent is not allowed to auto-run destructive operations; the metadata omission of required env vars should be corrected by the author before trusting automated workflows.

Type: OpenClaw Skill Name: tencentcloud-cvm-skill Version: 1.0.2 The skill is classified as suspicious due to the use of `sshpass` for password-based SSH authentication and the disabling of SSH host key checking (`StrictHostKeyChecking=no`, `UserKnownHostsFile=/dev/null`) in `scripts/common.sh` and various `scripts/ops/*.sh` files. While these are risky, they are explicitly documented as part of an O&M toolset. The skill also stores instance passwords locally in `~/.tencent_cvm_passwords` (with `chmod 600`), which is sensitive. However, the `SKILL.md` explicitly states that 'write operations' require 'manual confirmation', and `scripts/ops/remote-exec.sh` implements a robust whitelist/blacklist to prevent arbitrary command execution, command chaining, and remote payload execution, indicating an intent to control and limit potentially harmful actions rather than enable them maliciously. There is no evidence of intentional data exfiltration to external endpoints, persistence mechanisms, or prompt injection against the agent.