The Agent Skills Directory

[COMMAND_EXECUTION]: The skill operates by invoking the qodo CLI to manage review standards. It includes specific, secure instructions for locating the binary on POSIX and Windows systems and mandates the use of JSON flags to ensure that the agent parses output in a structured and predictable manner.\n- [INDIRECT_PROMPT_INJECTION]: Because the skill generates rule updates based on conversation history, it contains an indirect prompt injection surface. This is mitigated through strict instructions that require the agent to present the final draft of any rule to the user and obtain explicit confirmation before calling the CLI's write commands.\n- [EXTERNAL_DOWNLOADS]: The skill references the official vendor download domain (get.qodo.ai) for obtaining CLI installers. It contains explicit security guardrails that prohibit the agent from automatically downloading or piping scripts into a shell, requiring instead that the user handles the installation with checksum verification.\n- [PRIVILEGE_ESCALATION]: The skill includes a provision for troubleshooting authentication failures by requesting a single, read-only diagnostic command outside the execution sandbox. This is carefully controlled by requiring explicit user approval and prohibiting any persistent or automatic bypass of the sandbox for subsequent operations.

qodo-manage-standards