Rizz Coach

AI texting and flirting coach that grades your messages, improves your texts, generates replies, simulates conversations, and creates shareable results.

This skill's code and prompts match the described 'texting coach' behavior and include explicit safety checks — that's good. However, package.json depends on the Anthropic SDK but the skill metadata does not declare any required API key or environment variable. Before installing or enabling this skill you should: 1) Confirm which API key or credential the skill expects (e.g., ANTHROPIC_API_KEY or platform-injected key) and where it will be stored. 2) Prefer creating a dedicated, limited-scope API key (not your main account key) and restrict its usage/costs. 3) Review the remaining (omitted) source files for any network calls or unexpected endpoints beyond the LLM client. 4) Consider running the skill in a sandbox or with network monitoring the first time to ensure it only calls the expected LLM endpoint and does not exfiltrate data. If the author intends the platform to inject LLM credentials, ask them to declare that in the skill metadata so you can evaluate it properly.

Type: OpenClaw Skill Name: rizz-coach Version: 1.0.0 The Rizz Coach skill is a well-structured AI agent designed for analyzing and generating texting/flirting advice. A review of the source code and SKILL.md instructions reveals no evidence of malicious intent, data exfiltration, or unauthorized system access. The skill includes a comprehensive safety layer in 'src/core/safety.ts' that uses regex patterns to proactively block harassment, coercion, stalking, and explicit content. All LLM interactions in 'src/core/prompts.ts' are governed by strict system prompts that reinforce these safety boundaries, and the code lacks any high-risk execution patterns like 'eval' or shell command injection.