The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data (coding rules) retrieved via the qodo CLI or the rules-search MCP tool. This creates an attack surface where instructions embedded within retrieved rules could attempt to influence agent behavior during implementation or code review.
Ingestion points: Output from the rules-search MCP tool and qodo read rules search CLI command as specified in SKILL.md.
Boundary markers: Instructions require the agent to preserve rule names, references, and severities, and to explain each rule in the context of the task, which helps maintain the integrity of the data.
Capability inventory: The skill is authorized to execute shell commands via the qodo CLI for rule retrieval, identity checks, and managed updates.
Sanitization: There are no explicit instructions to sanitize or escape the content of the retrieved rules before they are processed by the agent.
[COMMAND_EXECUTION]: The skill uses local shell commands to interact with the qodo CLI. This includes identity checks (qodo read whoami), rule searching (qodo read rules search), and version probes (qodo --version). It also includes specific workarounds for PowerShell and Git Bash environments to ensure consistent execution.
[SAFE]: The skill references the vendor's official domain (get.qodo.ai) for obtaining installers and provides a manual update procedure for the CLI and agents. It proactively warns against unsafe practices, such as piping remote installer scripts directly into a shell, recommending instead the use of checksum-pinned installers.

qodo-get-rules