lector.dev

What It Is

Lector is a self-hosted language-learning reader positioned somewhere between LingQ, Clozemaster, and an Anki-based workflow. It supports importing EPUB, Markdown, web articles, or directly pasted text. While reading, you can click words for translations, mark them as new/learning/known, and build up your own personal vocabulary database. Data is stored locally in SQLite and IndexedDB, and the official messaging emphasizes that there are no cloud accounts, telemetry, or sync services.

Core Features and Ecosystem

Its feature set is fairly complete: reading, vocabulary mining, around 2,900 Afrikaans-English Tatoeba sentence pairs, multiple-choice and typed cloze exercises, five-level SRS intervals, AI correction for writing journals, an LLM Tutor, statistics dashboards, and optional TTS. AI Provider support includes Anthropic Claude, Apfel for Apple Intelligence, and local models via Ollama. If no API key is provided, it can fall back to a local dictionary of 2,000 high-frequency words. AnkiConnect integration is a highlight: it can push cards directly from the browser to a local Anki Desktop instance, though this requires Anki and the browser to run on the same machine/environment.

Deployment, Pricing, and Documentation

Lector is distributed as a Docker image, with Docker Compose recommended. You map ./data to persist databases and books. The documentation covers updates, backups, Tailscale/WireGuard access, Caddy reverse-proxy HTTPS, environment variables, and a complete compose example; overall, the documentation quality is high. The product itself is free, with no subscription or feature gates. However, Claude API usage, Claude Pro/Team quotas, and Google Cloud TTS are external costs.

Pros, Cons, and Who It’s For

The main advantages are its privacy-friendly design, offline-first approach, complete learning loop, and the option to use local LLMs to reduce reliance on cloud services. The downsides are also clear: the currently mature language pack is mainly Afrikaans→English, while Spanish and German are not yet available; translation quality from small models is limited; and deployment, reverse proxying, VPN access, and AnkiConnect configuration are not exactly easy for non-technical users. It is better suited to language learners who are comfortable with Docker, Anki, and self-hosting, rather than casual users who just want something that works out of the box.

Access from China

The source material does not provide information about accessibility from mainland China. Since it can be self-hosted, the core application can be used on a local network after deployment. However, external dependencies such as Anthropic, Google Cloud, GitHub Container Registry, and Tailscale may involve uncertainty around network access and payment. To reduce risk, users may want to prioritize local options such as Ollama or Apfel, and use local tools like Anki as alternatives or supplements.