Brain Tease Game

What to consider before installing: - Core functionality: The game runs locally using bundled question JSON files and stores sessions/history at ~/.cache/brain-teaser — this behavior matches the description. - AI generation: An optional AI generator can call LLM APIs if configured. By default AI generation is disabled, but the skill will try to discover credentials in this order: BRAIN_TEASER_API_KEY (env), ~/.claude/settings.json (Anthropic token), and ~/.openclaw/openclaw.json (local gateway). If any of those exist, the skill may use those credentials to make network calls. If you do not want this, do not set BRAIN_TEASER_API_KEY and ensure ~/.claude/settings.json and ~/.openclaw/openclaw.json do not contain sensitive tokens on that machine. - Dependency note: README/SKILL.md claim "no external dependencies (standard library only)", but ai_generator attempts to import the openai client if available. AI generation will only work if such a library is installed, but this is an undocumented optional dependency — expect that enabling AI may require installing an external Python package. - Risk summary: The main risk is accidental use of existing LLM credentials (or reading unrelated config files) and then making network requests. If you plan to use only the offline game, either run it in an environment without the LLM credentials present or audit/remove the AI-related files/env vars. Otherwise the skill appears consistent with its stated purpose. Suggested actions: - If you want the offline game only: do not set BRAIN_TEASER_API_KEY and run in an environment without ~/.claude/settings.json or ~/.openclaw/openclaw.json containing tokens. - If you want AI generation: review ai_generator.py to confirm which files/vars it reads and what endpoints it will call; consider creating a dedicated API key with limited scope. - Optional: run the skill in a sandbox or inspect network traffic the first time you enable AI to confirm behavior.