The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local shell commands to build application artifacts, interact with Git repositories, and manage Volcengine cloud resources using the ve CLI and other development tools (Docker, Kubectl).
[DYNAMIC_EXECUTION]: The utility script scripts/poll-status.sh uses the eval command to execute string-based probes passed as arguments. This is used internally to poll the status of cloud resources during the provisioning process.
[REMOTE_CODE_EXECUTION]: The skill automates the execution of commands on remote ECS instances using Volcengine's Cloud Assistant (RunCommand). This mechanism is used to install runtimes, upload application code, and configure systemd services as part of the intended deployment flow.
[DATA_EXFILTRATION]: The skill performs an outbound IP detection check specifically to restrict SSH access to the user's current network location. This is a standard security hardening practice for cloud infrastructure management.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from user-provided Git repositories (such as configuration files and dependency manifests) to generate deployment logic. To mitigate potential injection risks, it generates a human-readable .volcengine/deployment-plan.md which acts as a mandatory validation and approval gate before any resources are modified.

volcengine-deploy