The Agent Skills Directory

[PROMPT_INJECTION]: Indirect Prompt Injection Surface
Ingestion points: The skill ingests untrusted data from project source files using Read, Glob, and Grep tools while executing in Review mode and Audit mode.
Boundary markers: There are no defined delimiters or explicit warnings to ignore embedded instructions within the processed code, which could lead the agent to follow malicious prompts hidden in the files being audited.
Capability inventory: The skill possesses significant capabilities, including Bash for command execution, Write for file modification, and the Agent tool for delegating tasks to sub-agents. These tools could be leveraged by an attacker if the agent obeys instructions found in the code.
Sanitization: No validation or sanitization routines are specified for the content extracted from external files before it is processed or passed to subsequent agent steps.

golang-error-handling