The Agent Skills Directory

[SAFE]: The skill provides structured workflows for software architecture and technical planning. It emphasizes minimal changes, trade-off analysis, and documentation in a standard directory structure.
[COMMAND_EXECUTION]: The skill instructions include usage of bash for code analysis tools such as grep, find, and rg. These commands are used appropriately for local codebase searching and do not present a security risk as they are read-only search operations scoped to the project source.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface by reading existing code files to identify patterns. While this is a theoretical entry point for indirect instructions, the skill implements strong mitigation by requiring explicit user validation (Step 7) and approval of all generated designs before finalization.
[METADATA_POISONING]: The skill configuration includes after_complete hooks for session logging and agent improvement. These are standard platform telemetry features used for session management and do not perform unauthorized network operations or data exfiltration.

architecting-solutions