Ardor：能记住求职目标的 AI 职业智能体 v0.5.0 自托管预览版发布
isCharles/project-ardor
Ardor 是一款可自托管的 AI 职业智能体，v0.5.0 预览版现已发布，Windows 用户可从公开桌面版本下载安装包。它通过可见、可编辑的长期记忆串联简历分析、文字与实时语音面试练习、面试复盘、闪卡学习和日历计划，数据保留在用户自己的部署环境中。当前安装包未签名，技术文档暂为中文。
English · 简体中文
Your career shouldn't restart with every new conversation.
An AI career agent that remembers your goals, sees your progress, and helps you practice until you're ready.
Meet Ardor · What it does · Get started
The hard part isn't just finding the right answer
You revise your résumé and prepare for interview after interview. Yet every new tool asks you to explain your background, target role, and last weak answer all over again.
Ardor connects the work scattered across your résumé, interviews, notes, and calendar into a process you can actually continue. Talk to the agent or open a module directly. It can help you analyze, practice, and plan, while keeping you in control of consequential actions.
One conversation, real progress
“I have a Java backend interview next week. Here are my résumé and notes from my last interview. What should I work on first?”
Ardor can use what it already knows to turn that question into concrete preparation:
| Understand where you stand | Practice the real thing | Find the weak spots | Keep improving |
|---|---|---|---|
| Analyze your résumé and target role with structured recommendations | Practice project, technical, behavioral, and coding questions by text or voice | Organize past interviews question by question and identify gaps in your answers | Schedule learning, flashcards, and calendar tasks for what comes next |
The work does not disappear when the chat ends. Your documents, conversations, interviews, and plans stay in your own workspace.
Built for the real job search
✦ An agent with contextMultiple conversations and a visible, editable long-term memory. Ardor can reference your résumé, interview notes, knowledge documents, and learning plans, and ask when information is missing. |
✦ Résumé reports worth readingUpload a PDF or DOCX for asynchronous analysis. Reports present a score, strengths, and actionable improvements; previous uploads and reports remain available. |
✦ Interviews you can practiceWork through project deep dives and coding questions one at a time. Text and real-time voice interviews support pause detection and interrupting the AI interviewer. |
✦ Real interviews that teach you somethingTurn transcripts or recollections into question-by-question recaps, locate weak answers, and decide whether to create a focused improvement plan. |
✦ From “I don't know” to “I've practiced it”Generate explanations and exercises for weak concepts. Review flashcards on a schedule and see learning plans alongside tasks in one calendar. |
✦ Your knowledge, ready when you need itUpload material into a personal knowledge base or ask the agent to research online. Retrieved context can support later conversations instead of sitting in a folder. |
For job seekers—and people who want control of their tools
- One entry point, many paths. Start with chat or go straight to résumés, recaps, interviews, flashcards, learning, knowledge, and your calendar.
- Your choice of model configuration. An administrator can provide shared AI and voice services; individuals can also configure their own APIs.
- You confirm important actions. The agent may propose a deletion, but you approve data removal in the UI.
- Self-hostable today. Ardor currently runs on your own deployment, where your documents and data remain.
Get started
v0.5.0 is a self-hosted preview, not a public hosted service. Windows users can download the installer from public desktop releases. For local use, start the Ardor services with the setup guide, then launch the desktop shortcut. The current installer is unsigned, so Windows may warn that the publisher is unknown. You can also use http://localhost:3000 in a browser.
Explore the product scope, architecture, roadmap, and version policy. The desktop guide covers installation, server connections, and updates. Technical documentation is currently in Chinese.
Project Ardor · Keep the fire going.
来源：GitHub · github.com