The Agent Skills Directory

[COMMAND_EXECUTION]: The skill involves executing shell commands through the rtk wrapper tool. Examples include rtk git status, rtk git diff, and rtk grep. It also mentions rtk test <cmd>, where <cmd> represents arbitrary shell commands passed as arguments.
[INDIRECT_PROMPT_INJECTION]: The skill processes terminal output from external commands which may contain untrusted data. Instructions or malicious patterns embedded in logs, test results, or source code being inspected could influence the agent's subsequent actions.
Ingestion points: Terminal output from commands like git, grep, and test (SKILL.md).
Boundary markers: The skill does not define specific delimiters for command output, though the rtk tool itself acts as a processing boundary.
Capability inventory: Full shell access to run local commands and utilities (SKILL.md).
Sanitization: The skill relies on rtk filtering but warns that details can be omitted, recommending a switch to raw output for high-fidelity tasks.

rtk