v1.0.1 — Apache 2.0 licensed project code

This release licenses the original LDS project code under Apache 2.0. Both learner archives include LICENSE, NOTICE and LICENSING.md, and package metadata identifies Apache-2.0. Third-party components and provider services retain their own terms.

Choose the starter to implement the learning tasks, or the completed reference to inspect the finished application. Bring your own API keys. No private environment files or personal recordings are bundled. The accompanying SHA256SUMS.txt verifies both ZIPs.

Existing releases are unchanged. This packaging update does not reset LDS lesson progress.

The Chrome extension remains version 1.0.0; 1.0.1 is the licensed source-package release. All 29 offline tests pass.

Feed Lens 1.0

Feed Lens is an LDS guided project: build a Chrome extension that recommends what to read on LinkedIn Home using your own criteria and TypeSafe Jev key.

Choose your download

• Starter ZIP: implement the three questions and reading policy, with recovery checkpoints.
• Complete reference ZIP: run the finished application and explore the code.
• Source code: the same complete implementation, available to clone.

Start locally

Use Node.js 22.9+ and desktop Chrome. Extract the archive, run npm run check and npm start, and open http://127.0.0.1:3075. Follow LEARNER-GUIDE.md for all eight steps. There are no runtime dependencies to install. The included practice feed and offline tests need no API key. Live analysis uses your own TypeSafe account.

API keys stay in your private .env.local, never in extension files. The API-key template in this release is blank. LinkedIn Home is the supported live platform; Facebook, X and Reddit need separate integrations.

The default session cap is 40 attempted live calls. Set FEED_LENS_MAX_CALLS=unlimited in your private environment file to remove that cap deliberately; rate and concurrency limits still apply.

Verification: 29 passing standalone tests, fresh archive/recovery checks, and nine isolated browser integration groups. These checks establish application behavior, not model accuracy.