The Agent Skills Directory

[SAFE]: The skill is designed to create and maintain a local knowledge base for tracking decision-making processes. It uses a structured directory approach (facts, patterns, snapshots, and pending issues).
[COMMAND_EXECUTION]: The skill provides instructions for managing local files and directories (creating folder skeletons, reading status files, and writing Markdown updates). By default, it stores data in ~/.dbs/decisions/ or the current working directory. These operations are standard for a file-management skill and do not include arbitrary or dangerous system commands.
[DATA_EXFILTRATION]: There are no network-related commands (such as curl or wget) or indications that data is sent to external servers. The skill operates entirely on the local file system.
[PROMPT_INJECTION]: The instructions focus on organizational rules and tagging systems (e.g., [AI 推测], [AI 结论]) to ensure clarity between AI and user input. No adversarial prompts or safety bypass attempts were detected.
[CREDENTIALS_UNSAFE]: The skill does not access sensitive credential files (like .ssh or .env) and includes a "Privacy Mode" feature that encourages the use of aliases for sensitive identifiers, which is a security best practice.

dbs-decision