The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill documentation instructs users to install the 'orbit' CLI using the command curl -sSfL https://raw.githubusercontent.com/jorgemuza/orbit/main/install.sh | sh. This pattern downloads and executes a remote script with the user's privileges without any prior verification or checksum validation, which is a high-risk practice for remote code execution.
[COMMAND_EXECUTION]: The skill's core functionality is built around the execution of the external 'orbit' CLI tool. The agent is instructed to run various shell commands (e.g., orbit -p <profile> gl ...) to interact with GitLab services and local configuration files.
[PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection because it fetches and processes content from external GitLab resources such as merge requests, issues, and comments. 1. Ingestion points: Content from GitLab API retrieved via commands like gl mr view, gl mr notes, and gl issue view (SKILL.md, references/commands.md). 2. Boundary markers: No delimiters or specific instructions are provided to the agent to ignore potentially malicious content within the fetched data. 3. Capability inventory: The skill can execute shell commands via the 'orbit' CLI, access the network to call GitLab APIs, and read the user's local configuration at ~/.config/orbit/config.yaml. 4. Sanitization: No sanitization, escaping, or validation of the remote content is implemented before it is processed by the agent.

gitlab