Skip to content

LLM Wiki is a cross-platform desktop application that turns your documents into an organized, interlinked knowledge base — automatically. Instead of traditional RAG (retrieve-and-answer from scratch every time), the LLM incrementally builds and maintains a persistent wiki from your sources。

TypeScript

feat: configurable context window size with preset slider (4K → 1M)

nash_su9785ed9975d2
diff --git a/src/components/chat/chat-panel.tsx b/src/components/chat/chat-panel.tsx
index 63ae0217..b026402c 100644
--- a/src/components/chat/chat-panel.tsx
+++ b/src/components/chat/chat-panel.tsx
@@ -88,8 +88,9 @@ export function ChatPanel() {
graphExpansions.sort((a, b) => b.relevance - a.relevance)
// --- Phase 3 & 4: Budget control with priority queue ---
- const TOTAL_BUDGET = 30_000
- const MAX_PAGE_SIZE = 8_000
+ // Use ~60% of max context for page content, rest for prompt + history + response
+ const TOTAL_BUDGET = Math.floor(llmConfig.maxContextSize * 0.6)
+ const MAX_PAGE_SIZE = Math.min(Math.floor(TOTAL_BUDGET * 0.3), 30_000)
let usedChars = 0
type PageEntry = { title: string; path: string; content: string; priority: number }
diff --git a/src/components/settings/settings-view.tsx b/src/components/settings/settings-view.tsx
index eed5f8bc..87ab1c36 100644
--- a/src/components/settings/settings-view.tsx
+++ b/src/components/settings/settings-view.tsx
@@ -32,6 +32,7 @@ export function SettingsView() {
const [model, setModel] = useState(llmConfig.model)
const [ollamaUrl, setOllamaUrl] = useState(llmConfig.ollamaUrl)
const [customEndpoint, setCustomEndpoint] = useState(llmConfig.customEndpoint)
+ const [maxContextSize, setMaxContextSize] = useState(llmConfig.maxContextSize ?? 204800)
const [searchProvider, setSearchProvider] = useState(searchApiConfig.provider)
const [searchApiKey, setSearchApiKey] = useState(searchApiConfig.apiKey)
const [saved, setSaved] = useState(false)
@@ -54,7 +55,7 @@ export function SettingsView() {
async function handleSave() {
const { saveLlmConfig, saveSearchApiConfig } = await import("@/lib/project-store")
- const newConfig = { provider, apiKey, model, ollamaUrl, customEndpoint }
+ const newConfig = { provider, apiKey, model, ollamaUrl, customEndpoint, maxContextSize }
const newSearchConfig = { provider: searchProvider, apiKey: searchApiKey }
setSearchApiConfig(newSearchConfig)
await saveSearchApiConfig(newSearchConfig)
@@ -203,6 +204,18 @@ export function SettingsView() {
</div>
</div>
+ {/* Context Window Size */}
+ <div className="space-y-4 rounded-lg border p-4">
+ <h3 className="font-semibold">Context Window</h3>
+ <p className="text-xs text-muted-foreground">
+ Maximum context size sent to the LLM. Larger context allows more wiki pages in each query but costs more tokens.
+ </p>
+
+ <div className="space-y-3">
+ <ContextSizeSelector value={maxContextSize} onChange={setMaxContextSize} />
+ </div>
+ </div>
+
{/* Web Search API section */}
<div className="space-y-4 rounded-lg border p-4">
<h3 className="font-semibold">Web Search (Deep Research)</h3>
@@ -254,3 +267,65 @@ export function SettingsView() {
</div>
)
}
+
+// Context size presets matching common model context windows
+const CONTEXT_PRESETS = [
+ { value: 4096, label: "4K" },
+ { value: 8192, label: "8K" },
+ { value: 16384, label: "16K" },
+ { value: 32768, label: "32K" },
+ { value: 65536, label: "64K" },
+ { value: 131072, label: "128K" },
+ { value: 204800, label: "200K" },
+ { value: 262144, label: "256K" },
+ { value: 524288, label: "512K" },
+ { value: 1000000, label: "1M" },
+]
+
+function ContextSizeSelector({ value, onChange }: { value: number; onChange: (v: number) => void }) {
+ // Find closest preset index
+ const closestIndex = CONTEXT_PRESETS.reduce((best, preset, i) => {
+ return Math.abs(preset.value - value) < Math.abs(CONTEXT_PRESETS[best].value - value) ? i : best
+ }, 0)
+
+ return (
+ <div>
+ <div className="flex items-center justify-between mb-2">
+ <span className="text-sm font-medium">{formatSize(value)}</span>
+ <span className="text-xs text-muted-foreground">
+ ~{Math.floor(value * 0.6 / 1000)}K chars for wiki content
+ </span>
+ </div>
+ <input
+ type="range"
+ min={0}
+ max={CONTEXT_PRESETS.length - 1}
+ step={1}
+ value={closestIndex}
+ onChange={(e) => onChange(CONTEXT_PRESETS[parseInt(e.target.value)].value)}
+ className="w-full h-2 rounded-lg appearance-none cursor-pointer accent-primary"
+ style={{ background: `linear-gradient(to right, #4f46e5 ${(closestIndex / (CONTEXT_PRESETS.length - 1)) * 100}%, #e5e7eb ${(closestIndex / (CONTEXT_PRESETS.length - 1)) * 100}%)` }}
+ />
+ <div className="flex justify-between mt-1">
+ {CONTEXT_PRESETS.map((preset, i) => (
+ <button
+ key={preset.value}
+ type="button"
+ onClick={() => onChange(preset.value)}
+ className={`text-[9px] px-0.5 ${
+ i === closestIndex ? "text-primary font-bold" : "text-muted-foreground/50"
+ }`}
+ >
+ {preset.label}
+ </button>
+ ))}
+ </div>
+ </div>
+ )
+}
+
+function formatSize(chars: number): string {
+ if (chars >= 1000000) return `${(chars / 1000000).toFixed(1)}M characters`
+ if (chars >= 1000) return `${Math.round(chars / 1000)}K characters`
+ return `${chars} characters`
+}
diff --git a/src/stores/wiki-store.ts b/src/stores/wiki-store.ts
index 3ee3d669..275038f1 100644
--- a/src/stores/wiki-store.ts
+++ b/src/stores/wiki-store.ts
@@ -7,6 +7,7 @@ interface LlmConfig {
model: string
ollamaUrl: string
customEndpoint: string
+ maxContextSize: number // max context window in characters
}
interface SearchApiConfig {
@@ -46,6 +47,7 @@ export const useWikiStore = create<WikiState>((set) => ({
llmConfig: {
provider: "openai",
apiKey: "",
+ maxContextSize: 204800,
model: "",
ollamaUrl: "http://localhost:11434",
customEndpoint: "",