Security audit
Security checks for vulnerabilities and agentic risk
This Markdown-only Azure DevOps skill is not overtly malicious, but it requests write and command authority while giving mismatched, under-scoped instructions for impactful DevOps actions.
Review this skill before installing. Use only least-privilege Azure DevOps credentials, require the agent to show exact target projects, repositories, branches, pull requests, and work items before any create, merge, or update action, and do not rely on the provided examples as an accurate operation contract.
The skill supports impactful operations such as creating pull requests and managing work items, yet the documentation does not require an explicit user confirmation or warning before write/execute actions. In a DevOps environment, silent state-changing actions can modify repositories or workflows, causing operational disruption, unauthorized changes, or compliance issues.
The documented output schema describes a generic grading/reporting tool rather than Azure DevOps operations such as listing projects, branches, or creating pull requests. This semantic mismatch can mislead an agent or user about what actions the skill actually performs, increasing the risk of unintended execution paths, unsafe automation, or trust in incorrect results.
The examples show generic content-evaluation inputs and score-based outputs instead of Azure DevOps API operations. In an agent setting, misleading examples can cause the model to invoke the skill with wrong parameters or interpret unrelated outputs as successful DevOps actions, which is dangerous when the skill also has write/exec capabilities.
The markdown mentions API keys and cloud dependencies but lacks a clear privacy/data-handling warning for sending data to external Azure DevOps or related services. Users may unintentionally expose repository metadata, branch names, work-item contents, or tokens without understanding the network and retention implications.
The feature list claims broad file processing and system command execution capabilities beyond the stated Azure DevOps scope. Overbroad capability claims are risky because they expand the apparent trust boundary and may encourage an agent to use this skill for filesystem or shell tasks unrelated to DevOps, increasing the chance of abuse or privilege misuse.
No suspicious patterns detected.
