The Agent Skills Directory

[SAFE]: No malicious patterns detected. The skill follows best practices for platform integration and includes robust internal security guidelines for the agent.- [PROMPT_INJECTION]: The skill implements strong boundary instructions, specifically the 'Writing Operation Security Rules' in SKILL.md, which require explicit user confirmation for all data-modifying actions. This effectively mitigates the risk of indirect prompt injection from processed knowledge base content. Evidence: '核心原则：写入、修改、删除操作必须基于用户明确提供的目标信息，禁止 Agent 自行选择或猜测目标' in SKILL.md.- [DATA_EXFILTRATION]: Configuration management for access tokens is handled via standard local configuration files (~/.mcporter/mcporter.json) as described in setup.md. Network operations are restricted to the official service domains (lexiangla.com and lexiang-app.com), which are associated with the trusted Lexiang service.- [COMMAND_EXECUTION]: The skill includes local helper scripts (sync-folder.ts, upload-files.py) for file management. These scripts are well-documented, include unit tests (test_upload_files.py), and perform actions strictly within the scope of the skill's purpose. A self-maintenance mechanism in references/skill-maintenance.md allows the agent to update its own instructions upon explicit user request.

lexiang-knowledge-base