Security audit
Security checks for vulnerabilities and agentic risk
The skill is a coherent TDD helper, but it can automatically change linked GitHub issues and SPEC status, so it should be reviewed before installation.
Install this only if you want the agent to both implement approved SPECs and manage linked GitHub issue state. Use a GitHub token with limited repository permissions, verify the SPEC ticket points to the intended issue, and consider requiring explicit confirmation before label edits, comments, or issue closure.
The skill instructs the agent to read local SPEC files and additional reference documents, but it does not declare any explicit tool scope such as allowed read-only paths. That creates an authority/visibility mismatch where the agent may access repository files beyond what a user would reasonably expect from the skill metadata.
The instruction mandates that all questions and explanations to the user be in Portuguese, which is a language/locale restriction. The file does not provide opt-in, user choice, or a clearly justified region-specific requirement, so this is a natural-language policy violation.
The skill is presented as a TDD implementation helper, but it also instructs the agent to resolve linked GitHub issues, mutate workflow state, and synchronize issue labels. This expands the operational scope from code/test execution into external project management actions, increasing the chance of unauthorized or unintended side effects outside the user's immediate request.
The skill directs the agent to change SPEC status to In implementation and later to Done, while also changing GitHub issue state, but it never prominently warns the user about these write side effects. Hidden state mutation is dangerous because users may invoke the skill expecting code generation only, yet it silently alters project-tracking artifacts and audit history.
These sections contain concrete gh issue edit and gh issue comment commands, including label changes, blocker comments, and issue closing behavior, none of which are necessary to implement tests and code. If followed automatically, the skill can modify external GitHub resources, create misleading workflow signals, or close/retag issues without explicit user approval.
