The Agent Skills Directory

[Indirect Prompt Injection] (HIGH): The skill is designed to ingest and analyze untrusted external data (source code from the local workspace) while possessing high-privilege capabilities that can modify the system or execute commands.
Ingestion points: The skill utilizes Read, Grep, and Glob tools to pull file contents into the agent's context for review.
Boundary markers: Absent. There are no instructions or markers defined to help the agent distinguish between its own instructions and potentially malicious instructions embedded in the code being audited.
Capability inventory: The agent is granted Bash, Write, Edit, and Task capabilities, which allow for arbitrary command execution and filesystem modification.
Sanitization: Absent. Source code is ingested and processed as raw text without validation or escaping.
[External Downloads] (MEDIUM): The workflow recommends the use of dotnet outdated, which is a third-party community tool for .NET. This creates a dependency on unverified external code that may be downloaded and executed if not already present.
[Command Execution] (LOW): The skill relies on Bash and Grep for searching through directories. While intended for auditing, these tools provide broad access to the host environment and file metadata.

security-review