hTangle/voice-interview-coach ↗★ 0
dsh-plugin-voice-interview-coach
🎙️ Voice Interview Coach — a DSH (DeepSeek Harness) plugin that runs a natural-language-graded, voice-based mock interview using the candidate's project, language, resume and target role, then produces a gap-analysis review with Markdown export and persistent memory.
AI Analysis
核心用途是进行语音模拟面试与复盘。适合需要针对特定岗位、项目和简历进行模拟问答，并获取 Markdown 格式差距分析报告的求职者。
Install
This plugin has no verified bundle, or compatibility checks failed. Read the repository notes first. Read the full README ↗
README
Read the full README ↗Usage
- Open the DSH session where the plugin is loaded. Run the
cordis_define/cordis_runcard (or the plugin's own entry point). - Fill in:
- Target role / difficulty — natural language
- Focus / style — what the interviewer should probe (optional)
- Question types — chip multi-select
- Project, programming language, resume (paste or upload)
- Question count (3–15)
- Click 开始语音面试. The interviewer speaks the first question.
- Click 🎙 语音回答 (or type in the textarea and press Enter) after each question.
- When the count is reached — or you click 结束并复盘 — the review appears.
- Use ⬇ 导出 Markdown to download it, 💾 写入记忆 to re-save it, and switch to the 历史对比 tab to compare multiple interviews.
Configuration
| Field | Where | Default | Description |
|---|---|---|---|
| Memory directory | Host | /.interview-memory/ | Where review Markdown files are stored |
| Resume size limit | Client | 5 MB upload, 20 000 chars after parsing | Larger resumes are truncated before being sent to the model |
| LLM provider | Host | first registered provider | Selected via ctx.llm.listProviders()[0]; change in code or configure your deployment's default |
| Speech language | Client | zh-CN | Change rec.lang / SpeechSynthesisUtterance.lang in src/client.js |