The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes ssh to perform administrative tasks on remote systems, including repository management and service verification.
[EXTERNAL_DOWNLOADS]: The skill clones provisioning tools from github.com/kylelundstedt/iv-provision.git and references container images from ghcr.io/kylelundstedt/exeslim-dev. These are authorized vendor resources belonging to the skill author.
[INDIRECT_PROMPT_INJECTION]: The skill instructions provide a surface for command injection by interpolating user-supplied strings directly into shell execution blocks.
Ingestion points: VM identifiers (<vm>), Git release tags (<tag>), and integration names are accepted from the user in SKILL.md.
Boundary markers: No delimiters or defensive instructions are provided to prevent the agent from executing malicious code embedded within these inputs.
Capability inventory: The agent is granted capabilities to execute ssh, git, curl, rm, and op (1Password CLI).
Sanitization: There are no instructions for the agent to sanitize, escape, or validate the user-provided inputs before their inclusion in shell commands.

upgrade-vm