up

up is a privacy-first desktop workspace for managing job applications, interview timelines, and source-backed job intelligence.

The public release starts completely empty. It contains no companies, positions, JD text, notes, timeline nodes, intelligence results, personal paths, or profile data.

Acknowledgements / 致谢

up is an independent visual companion built on and integrated with career-ops.

特别感谢 @santifer 创建并开源 career-ops。up 使用了 career-ops 提供的开源求职工作流、运行模式和本地数据契约， 在这些能力之上制作了桌面可视化界面，并继续加入适合个人求职使用的新功能。

career-ops provides the underlying job-search capabilities that up invokes, including role evaluation, opportunity scanning, tailored CV and cover-letter workflows, application preparation, tracking, follow-up, and interview preparation. up connects to a separately installed local copy of career-ops and makes those workflows easier to see and operate from a desktop interface.

This project adds its own:

• Visual Career Ops action entry points inside a macOS desktop app
• Company-centered position, JD, status, and intelligence workspace
• Editable application Gantt timeline and a separate sourced public-process timeline
• Source-linked recruiting intelligence, experience research, and update center
• Local-first data handling, visual company presets, and a human confirmation boundary

up is a community project maintained independently by @differance-dfhs. It is not an official career-ops release and does not claim endorsement, sponsorship, or affiliation with career-ops or its maintainer. The upstream project remains available under its own MIT license and trademark policy. See Third-party notices.

What it does

• Company-centered overview with circular company clusters and role switching
• Safe deletion for either the current role or an entire company cluster
• Directly editable Gantt timeline
• Separate public recruiting-process timeline with exact official dates, evidence-based historical ranges, and confidence labels
• Application progress derived from dated nodes
• Position-level JD and notes
• Intelligence reader for market opportunities and source-linked interview preparation
• Expandable interview experience library organized by background, process, interview round, question, answer, and original post
• Autumn-recruiting full-time opportunity filter that excludes internships and unverified social-recruiting roles
• In-app Loop update center with unread indicators and a retained change feed
• Optional daily Codex Loop integration

Install on macOS

Download the latest .dmg from GitHub Releases, open it, and drag up to Applications.

The current local package is built for Apple Silicon. The release workflow also supports Intel builds.

Because the app is distributed without an Apple Developer ID signature, macOS may require: right-click the app, choose Open, then confirm once.

Privacy

• All workspace data is stored locally.
• The app does not include analytics, accounts, or a remote database.
• Installed data lives in ~/Library/Application Support/up/data.
• The GitHub repository contains only empty JSON templates.
• Codex integration is opt-in and must be created by the user.

Connect Codex

1. Install and open up.
2. Copy the automation prompt from docs/codex-loop-prompt.md.
3. Ask Codex to create the daily 22:30 automation.
4. Use ~/Library/Application Support/up/data as {{UP_DATA_DIR}}.

Before each run, Codex reads the latest positions from workspace.json. It searches official sources and publicly accessible experience posts, including relevant Xiaohongshu posts from current and earlier cohorts. It writes sourced opportunities and structured interview preparation to intelligence.json. The workbench checks for updates while it is open.

Development

npm install
npm run dev

Build and validate:

npm run build
npm run test:sites

If you are deploying through OpenAI Sites and have a local .openai/hosting.json, prepare the Sites bundle with:

npm run build:sites

Build a macOS Apple Silicon installer:

npm run desktop:dist

Build an Intel installer:

npm run desktop:dist:x64

Data format

• data/workspace.json: local companies, positions, JD, notes, and timeline nodes
• data/intelligence.json: Codex-generated opportunities and position briefs

The committed versions of both files are intentionally empty.

License

MIT

Third-party font and trademark notices are documented in THIRD_PARTY_NOTICES.md.