review simulate

进行中文求职语音模拟面试。适用于用户想围绕目标岗位进行多轮中文面试练习，并通过 ASR 识别回答、由 LLM 决定追问或换题、由 TTS 播报面试官问题，最终输出结构化评估报告、改进建议和示例优化回答的场景。支持 `target_role`、`interviewer_style`、`min_rounds`、`ma...

Before installing or running this skill, consider the following: - The metadata claims no credentials, but the script requires API keys for an LLM and for ASR/TTS; you will need to provide INTERVIEW_LLM_API_KEY (or IME_MODEL_API_KEY) and INTERVIEW_ASR_API_KEY / INTERVIEW_TTS_API_KEY (or SENSEAUDIO_API_KEY). The skill will exit if those keys are missing. - The default endpoints in code point to models.audiozen.cn and api.senseaudio.cn — if you don't recognize those services, verify their privacy and billing policies before supplying keys or use your own endpoints/keys. - The script uploads audio and text to external services (ASR/TTS/LLM). If you will be processing sensitive audio (personal data, interviews containing private info), avoid sending it to third parties or sanitize it first. - The script loads .env from the skill directory and the current working directory; be careful that running it in a directory with other .env files doesn't leak unrelated secrets. Inspect the .env.example referenced in SKILL.md (the path in the doc looks like another workspace) and confirm it is appropriate. - If you want to audit behavior, review the full scripts/run_interview.py to confirm no unexpected network calls beyond the three service endpoints and to confirm where outputs are written (outputs/final_report.json and audio files). - If you prefer not to use external services, modify the code to use local/text-only mode (the script supports typing text instead of audio) or substitute endpoints with services you control. Given the clear manifest vs code mismatch about required credentials, treat this skill as suspicious until you confirm the endpoints and provide only credentials you are comfortable using with those services.

Type: OpenClaw Skill Name: review-simulate Version: 1.0.0 The skill bundle is a functional Chinese voice interview simulator that uses LLM, ASR, and TTS services. The Python script (scripts/run_interview.py) and associated prompts (references/prompts_cn.md) are well-structured and strictly follow the logic described in SKILL.md. There is no evidence of data exfiltration, malicious execution, or harmful prompt injection; the code uses standard libraries (OpenAI, requests) and handles API keys via environment variables as expected for this type of application.