The Agent Skills Directory

Diagnostic Patterns for Risk Detection: The skill contains references to sensitive file paths (such as ~/.aws/credentials or ~/.ssh/) and potentially risky code patterns (like eval or subprocess.run). These references are part of a security checklist used by the agent to scan the target codebase for vulnerabilities. They represent the skill's diagnostic capabilities and do not indicate that the skill itself is accessing these locations or executing such commands.
Controlled Output Schema: The skill is instructed to output its findings into a structured JSON format at a predefined location. This design ensures that the diagnostic results are handled in a predictable manner by the calling CLI tool, minimizing the risk of arbitrary data exposure.
Read-Only Scope: The instructions explicitly constrain the agent to read-only analysis of the workspace. This is a best-practice for diagnostic tools, ensuring that the project being analyzed is not modified during the inspection process.
Supply Chain Security Focus: A significant portion of the skill (specifically supply-chain-checks.md) is dedicated to identifying sophisticated attack vectors like fetch-then-execute droppers or persistence mechanisms in third-party dependencies. This is a beneficial security feature provided to the end user.

azure-functions-doctor