The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The documentation in SKILL.md provides an installation command that pipes a remote script directly to the system shell: curl -fsSL https://raw.githubusercontent.com/rtk-ai/rtk/refs/heads/master/install.sh | sh. This facilitates unauthenticated, arbitrary code execution from an untrusted external repository.
[COMMAND_EXECUTION]: The scripts/rtk-wrapper.cjs helper script uses child_process.spawn to execute shell commands. This capability allows the agent to run any command on the host system via the rtk proxy, which could be abused to perform unauthorized actions or access sensitive data.
[EXTERNAL_DOWNLOADS]: The skill explicitly depends on an external binary (rtk) and recommends downloading it from a third-party GitHub repository (rtk-ai/rtk). This introduces supply chain risks as the source is not a verified or well-known organization.
[INDIRECT_PROMPT_INJECTION]: The skill processes output from shell commands, creating a vulnerability where malicious instructions in command output could influence the agent.
Ingestion points: Captured stdout/stderr from commands in scripts/rtk-wrapper.cjs reported to the agent.
Boundary markers: None identified in the workflow or output reporting logic.
Capability inventory: Full subprocess execution capability via spawn in the wrapper script.
Sanitization: No filtering or sanitization of captured command output is performed.

rtk-exec