The Agent Skills Directory

[SAFE]: The skill instructions define a legitimate and secure workflow for code and task review. It emphasizes independent verification, role separation (preventing the reviewer from editing the code), and empirical evidence collection.
[COMMAND_EXECUTION]: The skill mentions the use of a terminal tool to build, test, and run code. This is a standard and expected capability for a developer-oriented review skill. The instructions specify that the agent should run the project's existing tests and linting tools rather than arbitrary or hidden commands.
[DATA_EXFILTRATION]: No network exfiltration or unauthorized data access patterns were detected. The skill uses native Kanban tools (kanban_show, kanban_complete) and standard file reading tools for its primary functions.
[PROMPT_INJECTION]: No malicious instruction overrides, jailbreak attempts, or safety filter bypasses were found. The 'Instructions' are focused on quality assurance and process compliance.
[EXTERNAL_DOWNLOADS]: The skill does not perform any remote code downloads or package installations from external sources.

sdlc-review