Settings
All settings live under bitvanes.* in settings.json.
Walkthrough
| Setting | Default | Description |
|---|---|---|
bitvanes.walkthrough.style | standard | Audience tuning for AI walkthrough explanations: expert (terse, high-signal, risk-focused), standard (what + why), learner (plain language, syntax explained, jargon defined). Instant mode is unaffected |
Model
| Setting | Default | Description |
|---|---|---|
bitvanes.llm.provider | auto | auto (Copilot first, then the configured endpoint), vscode-lm, or openai-compatible |
bitvanes.llm.baseUrl | http://localhost:11434/v1 | OpenAI-compatible base URL. Ollama :11434/v1, LM Studio :1234/v1, vLLM :8000/v1, OpenAI https://api.openai.com/v1, Anthropic https://api.anthropic.com/v1, OpenRouter https://openrouter.ai/api/v1 |
bitvanes.llm.modelId | (empty) | Model ID for the endpoint, e.g. qwen2.5-coder:14b, gpt-4o, claude-sonnet-4-20250514. Ignored for Copilot |
bitvanes.llm.temperature | 0.2 | Low keeps schema-accurate JSON |
bitvanes.llm.maxTokens | 4096 | Max tokens requested from the model |
bitvanes.llm.maxSteps | 12 | Cap on walkthrough length (1–50) |
bitvanes.llm.contextLines | 6 | Context lines padded around each diff hunk in the prompt |
Playback
| Setting | Default | Description |
|---|---|---|
bitvanes.autoplay.intervalMs | 4000 | Delay between autoplay steps (min 1000); manual navigation pauses autoplay |
Editor
| Setting | Default | Description |
|---|---|---|
bitvanes.editor.spotlightColor | #f59e0b | Spotlight border color (hex) |
bitvanes.editor.dimOpacity | 0.3 | Opacity for code outside the active scope |
TIP
API keys are deliberately not settings — run BitVanes: Set API Key for External LLM Provider instead, which stores them in VS Code SecretStorage.