Automating Bilibili's hardest trivia gauntlet with LLMs

What it does

Bilibili reserves a special “hardcore member” badge for users who pass a brutal 100-question quiz—available only to level-6 accounts, three attempts per day. This tool automates the entire flow: it calls Bilibili’s API directly (no screen scraping or OCR), feeds each question to a configurable LLM, and submits the answer. The author recommends Silicon Flow’s API with a free ¥14 credit tier.

The interesting bit

The project is written in Rust but ships as a self-contained binary with musl static linking for Linux, universal binaries for macOS, and a one-liner install script. That’s unusually polished deployment hygiene for what is essentially a quiz-bot. The direct API approach also means it avoids the brittle screenshot-and-OCR pipelines common in similar tools.

Key highlights

• Direct Bilibili API integration, no OCR or browser automation
• Supports any OpenAI-compatible LLM endpoint via URL + model + API key
• Self-updating binary (bili-hardcore update) and clean uninstall
• Cross-platform releases: macOS universal, Windows x64, Linux x64/ARM64 (musl)
• Interactive or headless configuration modes

Caveats

• Requires a third-party LLM API key; no local model support mentioned
• Daily attempt limits (3) and 24-hour lockouts are enforced by Bilibili, not the tool
• Breaking config changes between 0.x and current versions require manual cleanup

Verdict

Worth a look if you’re a Bilibili regular who wants the hardcore badge without the trivia grind. Skip it if you don’t have a level-6 account or object to routing your quiz answers through a cloud LLM.