The Agent Skills Directory

[SAFE]: The skill provides static architectural documentation and decision-making templates. The content is purely educational and instructional, serving as a framework for human-led design decisions.- [NO_CODE]: There are no executable scripts, binaries, or command instructions included in the skill. The logic is handled through natural language instructions and templates.- [PROMPT_INJECTION]: The instructions do not contain attempts to override agent behavior, bypass safety filters, or extract system prompts. The role-play guidelines are strictly within the context of architectural analysis.- [COMMAND_EXECUTION]: A review of the included audit metadata (skill-report.json) confirms that earlier automated detections for command execution were false positives triggered by Markdown formatting (backticks used for file names). No actual shell commands or subprocess calls are present.- [EXTERNAL_DOWNLOADS]: The skill does not fetch remote scripts, external packages, or configuration files from the internet.- [DATA_EXFILTRATION]: No patterns for accessing sensitive local files (such as credentials or environment variables) or exfiltrating data via network requests were found.

architecture