The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the local repository and external report files, creating a surface for indirect prompt injection where malicious content in the source code could attempt to influence the agent's review output or behavior.
Ingestion points: The skill reads project source files from the src/ directory and several JSON audit reports (e.g., typescript.json, lint.json, deps.json) located in the .frontend-review/report/latest/raw/ path.
Boundary markers: There are no specific delimiters or "ignore embedded instructions" warnings defined in the SKILL.md to protect the agent from instructions embedded within the analyzed source code or JSON data.
Capability inventory: The skill uses the agent's capabilities to read arbitrary repository files, execute git log to identify modified files, and write a markdown report to the .frontend-review/report/latest/md/ directory.
Sanitization: The instructions do not provide methods for sanitizing, escaping, or validating the content of the project files or audit JSONs before they are processed by the agent.

frontend-expert