Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed engineering plan-review workflow that reads project context and may create narrow planning notes, without hidden execution, persistence, or credential behavior.
Install this if you want an opinionated plan-review assistant that will inspect repository context and produce planning artifacts. Before using it, be aware it may create a test-plan file and may ask to update TODOS.md; decline those options if you want a read-only review.
The skill encourages proactive invocation whenever a user has a plan or design doc and is about to start coding, which is a broad trigger surface. Over-broad auto-suggestion can lead to unintended activation of a high-instruction skill that performs repository inspection and proposes writes, increasing the chance of acting outside the user's actual intent.
The skill is presented as a plan review workflow, but it instructs the agent to create and write a test-plan artifact on disk. That expands behavior from analysis into state-changing repository modification, which can surprise users, create unwanted files, and be abused as a stepping stone for broader unsolicited workspace changes.
The skill goes beyond reviewing a plan by directing TODOs.md updates, which is a repository modification unrelated to a strictly interactive review. This can cause unauthorized scope expansion, pollute project documentation, and normalize write actions in contexts where the user asked only for analysis.
The skill's purpose is to review plans and architecture, but it instructs the agent to query PR metadata and repository defaults via gh and inspect git branch state. While possibly useful context, these repository/CLI interrogation capabilities are not clearly justified by the manifest's narrow eng-review description.
No suspicious patterns detected.
