The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process potentially untrusted data such as error logs, stack traces, and source code from external environments.
Ingestion points: Source files accessed via read_file, code patterns via search_files, and terminal output from tests or logs.
Boundary markers: The skill lacks specific delimiters or warnings to ignore instructions that might be embedded within the error messages or code comments it processes.
Capability inventory: The skill utilizes terminal for command execution, read_file/search_files for filesystem access, and delegate_task for multi-agent workflows.
Sanitization: There are no explicit instructions to sanitize or escape data gathered during the investigation phase before it is used to form hypotheses or implementation plans.

systematic-debugging