Security audit
Security checks for vulnerabilities and agentic risk
This skill persistently summarizes completed projects into reusable skill and memory files, which matches its stated purpose but requires careful review for sensitive details.
Install only if you want OpenClaw to help create persistent project-retrospective skills and memory files. Before approving saved output, review it for secrets, API keys, internal URLs, customer data, proprietary implementation details, and anything you would not want reused in future sessions.
The trigger phrases include very broad everyday language such as summary-style requests, which can cause the skill to activate in situations where the user did not specifically intend durable knowledge extraction or storage. In this skill, unintended activation is more dangerous because activation leads to analysis of long conversation history and possible persistence into skill/memory files.
The skill directs the agent to analyze lengthy conversation history and extract project goals, decisions, problems, and best practices into reusable artifacts without any minimization, redaction, or sensitivity review. That creates a real data retention risk: secrets, internal architecture, customer information, credentials, and proprietary workflows discussed during the project could be copied into persistent files and later exposed or reused out of context.
The memory template explicitly encourages recording detailed failed attempts, problem descriptions, and final solutions into persistent memory files. In practice, those sections often contain sensitive debugging traces, environment paths, operational mistakes, credentials handling details, or confidential business context, making the retention risk stronger in this skill than in a generic note-taking context.
The skill content, trigger examples, and user-facing confirmation prompt are all specified solely in Chinese, with no indication that the assistant should adapt to the user's preferred language. This can violate a language/locale policy if the organization expects language choice or user opt-in rather than forcing a single language.
No suspicious patterns detected.
