The Agent Skills Directory

[COMMAND_EXECUTION]: The skill acts as a proxy for executing a wide range of shell commands, including system-level operations for version control (Git), container management (Docker, Kubernetes), and various build/test runners.
[CREDENTIALS_UNSAFE]: Includes explicit instructions for the agent to access and filter environment variables using rtk env -f AWS, which is a common location for sensitive cloud credentials.
[DATA_EXFILTRATION]: The skill provides mechanisms to read local files (rtk read) and directory structures, facilitating the movement of potentially sensitive local code and data into the AI's context.
[PROMPT_INJECTION]: The skill processes untrusted data from several sources without defining safety boundaries, creating an attack surface for indirect prompt injection.
Ingestion points: Data enters the context via file reads (rtk read), network requests (rtk curl), and external tool outputs like git logs and container logs.
Boundary markers: No delimiters or instructions to ignore embedded commands are present in the provided instructions.
Capability inventory: The skill allows execution of arbitrary shell commands through the rtk wrapper across infrastructure and development toolsets.
Sanitization: No evidence of output sanitization or validation of the ingested content is provided.

rtk-compress