The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill executes a remote shell script from the vendor's GitHub repository using the pattern curl -fsSL https://raw.githubusercontent.com/opendeploy-dev/opendeploy-opencode/main/install.sh | bash. This occurs during the OpenCode plugin installation and update flows.
[COMMAND_EXECUTION]: The skill performs global system modifications, including installing global Node.js packages (npm install -g @opendeploydev/cli@latest) and using launchctl setenv to modify and persist environment variables like PATH and HTTP_PROXY on macOS. These changes affect the environment of all GUI applications launched by the user.
[EXTERNAL_DOWNLOADS]: Downloads software and configuration scripts from npmjs.org and github.com. It also suggests using registry.npmmirror.com as a fallback mirror for installations.
[PROMPT_INJECTION]: The skill contains an indirect injection surface by interpolating user-provided inputs into shell commands.
Ingestion points: User-provided strings for PATH and proxy <PORT> in the 'PATH repair' and 'Proxy repair' sections of SKILL.md.
Boundary markers: None; the skill directly instructs the agent to use the provided strings in commands.
Capability inventory: The skill has access to launchctl, npm, and the opendeploy CLI for system-level changes.
Sanitization: No validation or escaping is specified for the interpolated user strings before they are executed in launchctl setenv commands.

opendeploy-setup