The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data in the form of Python source code. Since the agent has access to the system shell through the Bash tool, a malicious actor could embed instructions within code comments or strings to attempt to manipulate the agent's output or execute unauthorized commands.
Ingestion points: The skill reads and processes Python files via the Read and Grep tools for code review purposes.
Boundary markers: There are no explicit instructions or delimiters defined to distinguish between the agent's instructions and the untrusted content being reviewed.
Capability inventory: The skill has access to the Bash tool, which can execute arbitrary shell commands.
Sanitization: The skill does not provide guidelines for the agent to sanitize or filter potential instructional content embedded within the source code it analyzes.

python-reviewer