通过 AIsa 执行网页、多源或近 30 天研究检索。触发条件：当用户需要搜索、研究、比对或趋势归纳时使用。支持多源检索与结构化输出。

This package is largely coherent with its stated purpose, but before installing or running it you should: 1) only provide a trusted AISA_API_KEY and avoid re-using highly privileged keys; 2) run it in an environment with Python 3.12+ and consider network isolation if you don't want it contacting external APIs; 3) be aware it will read/write a repo-local config file at .last30days-data/config.env (check file permissions and contents); 4) note the default Xiaohongshu helper base is host.docker.internal:18060 — the tool will probe that endpoint if present, which could reach local services; and 5) if you need higher assurance, review the included provider modules (scripts/lib/) for any unexpected outgoing endpoints or data handling before providing credentials.

Type: OpenClaw Skill Name: last30days-zh Version: 1.0.0 The last30days-zh skill bundle is a comprehensive research tool designed to aggregate and rank information from multiple sources including Reddit, X (Twitter), YouTube, and Polymarket. The code is well-structured and aligns with its stated purpose, utilizing the AIsa API (api.aisa.one) for hosted retrieval and standard public JSON endpoints for others. Notably, the tool includes security-conscious features such as checking for overly permissive file permissions on configuration files (scripts/lib/env.py) and using defensive 'untrusted_content' fencing in LLM prompts to mitigate prompt-injection risks from scraped data (scripts/lib/rerank.py). No malicious behaviors, such as unauthorized data exfiltration or persistence mechanisms, were detected.