The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill uses npx -y codereview-axi to download and execute a command-line utility from the NPM registry. This tool is a resource associated with the skill's author and is central to its primary function.- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from a remote server and uses it to guide the agent's actions. * Ingestion points: Data and comments returned by the cr wait command from the remote review service. * Boundary markers: None; the skill does not define delimiters or warnings to ignore embedded instructions in the comments. * Capability inventory: The agent can modify local files (fix verdict), perform network operations (cr reply, cr refresh), and execute instructions provided in the next_step field of the remote payload. * Sanitization: None; the agent is instructed to follow imperative instructions (next_step) and apply code changes directly based on quoted text.- [COMMAND_EXECUTION]: The skill executes the cr CLI tool with various arguments, including $ARGUMENTS provided by the user and instructions received from the remote service, which represents a potential command injection surface if the CLI tool does not provide adequate sanitization.- [DATA_EXFILTRATION]: The skill transmits local git diffs to an external web service to facilitate the browser-based review. While this is the intended purpose of the tool, it involves sending potentially sensitive repository content to a third-party server.

code-review