The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted project data (source code) through tools like read_file and search_code. This data is then used to perform sensitive operations such as write_file, edit_file, and run_tests. There are no explicit boundary markers or sanitization steps defined to prevent instructions embedded in the source code from overriding the agent's behavior.
Ingestion points: The skill reads project files and searches code using read_file and search_code in SKILL.md.
Boundary markers: No specific delimiters or warnings to ignore instructions within the processed data are defined.
Capability inventory: The agent has the capability to modify the codebase via write_file, edit_file, and create_file, and to execute code via run_tests (as listed in SKILL.md).
Sanitization: No sanitization or validation of the content read from files is performed before it is used in downstream operations.

develop-laravel-ecommerce-feature