Skip to content

Never stop coding. Free MIT AI gateway: one endpoint, 375 providers (150+ free), 1,400+ models Kimi, Claude, GPT, Gemini, GLM, DeepSeek, MiniMax. Works with Claude Code, Codex, Cursor, OpenCode, Cline & Copilot. Quota-aware auto-fallback, RTK+Caveman compression saves 15-95% tokens, MCP/A2A, Desktop/PWA. Built by 900+ contributors

Hide combo compression controls when disabled (#1840)

Randi3d70ad414ef9
diff --git a/open-sse/handlers/chatCore.ts b/open-sse/handlers/chatCore.ts
index ea09957f63..bbac5a2ae5 100644
--- a/open-sse/handlers/chatCore.ts
+++ b/open-sse/handlers/chatCore.ts
@@ -136,6 +136,7 @@ import {
isFallbackDecision,
EMERGENCY_FALLBACK_CONFIG,
} from "../services/emergencyFallback.ts";
+import type { CompressionConfig } from "../services/compression/types.ts";
import { prepareWebSearchFallbackBody } from "../services/webSearchFallback.ts";
import {
resolveExplicitStreamAlias,
@@ -1561,15 +1562,37 @@ export async function handleChatCore({
body?.messages || body?.input || body?.contents || body?.request?.contents || [];
if (body && Array.isArray(allMessages) && allMessages.length > 0) {
let estimatedTokens = estimateTokens(JSON.stringify(allMessages));
+ let promptCompressionEnabled = false;
+ let compressionSettings: CompressionConfig | null = null;
+
+ try {
+ const { getCompressionSettings } = await import("../../src/lib/db/compression.ts");
+ compressionSettings = await getCompressionSettings();
+ promptCompressionEnabled = compressionSettings.enabled;
+ } catch (err) {
+ log?.warn?.(
+ "COMPRESSION",
+ "Compression settings lookup skipped: " + (err instanceof Error ? err.message : String(err))
+ );
+ }
// --- Modular Compression Pipeline (Phase 1 Lite + Phase 2 Standard/Caveman + Phase 3 Aggressive) ---
// Runs BEFORE the existing reactive compressContext() to proactively reduce tokens.
try {
- const { getCompressionSettings } = await import("../../src/lib/db/compression.ts");
const { selectCompressionStrategy, applyCompression } =
await import("../services/compression/strategySelector.ts");
const { trackCompressionStats } = await import("../services/compression/stats.ts");
- let config = await getCompressionSettings();
+ let config: CompressionConfig = compressionSettings ?? {
+ enabled: false,
+ defaultMode: "off",
+ autoTriggerTokens: 0,
+ cacheMinutes: 5,
+ preserveSystemPrompt: true,
+ comboOverrides: {},
+ };
+ if (!promptCompressionEnabled || !compressionSettings) {
+ log?.debug?.("COMPRESSION", "Prompt compression disabled or unavailable");
+ }
let compressionComboKey = comboName ?? null;
if (isCombo && comboName) {
try {
@@ -1704,6 +1727,12 @@ export async function handleChatCore({
}
// --- End Modular Compression Pipeline ---
+ if (!promptCompressionEnabled) {
+ log?.debug?.(
+ "CONTEXT",
+ "Skipping proactive context compression: Prompt Compression disabled"
+ );
+ }
let contextLimit = getTokenLimit(provider, effectiveModel);
if (isCombo && comboName) {
@@ -1748,7 +1777,7 @@ export async function handleChatCore({
`Checking compression: ${estimatedTokens} tokens vs ${threshold} threshold (${contextLimit} limit, ${reservedTokens} reserved)`
);
- if (estimatedTokens > threshold) {
+ if (promptCompressionEnabled && estimatedTokens > threshold) {
log?.info?.(
"CONTEXT",
`Proactive compression triggered: ${estimatedTokens} tokens > ${threshold} threshold (${contextLimit} limit)`
diff --git a/src/app/(dashboard)/dashboard/combos/page.tsx b/src/app/(dashboard)/dashboard/combos/page.tsx
index d992b9f916..58c8b3ebca 100644
--- a/src/app/(dashboard)/dashboard/combos/page.tsx
+++ b/src/app/(dashboard)/dashboard/combos/page.tsx
@@ -606,6 +606,7 @@ export default function CombosPage() {
const [comboDragOverIndex, setComboDragOverIndex] = useState(null);
const [savingComboOrder, setSavingComboOrder] = useState(false);
const [comboConfigMode, setComboConfigMode] = useState("guided");
+ const [promptCompressionEnabled, setPromptCompressionEnabled] = useState(false);
const [selectedIntelligentComboId, setSelectedIntelligentComboId] = useState<string | null>(null);
const comboDragIndexRef = useRef<number | null>(null);
const activeFilter = normalizeIntelligentRoutingFilter(searchParams.get("filter"));
@@ -650,6 +651,10 @@ export default function CombosPage() {
.then((r) => (r.ok ? r.json() : null))
.then((settings) => setComboConfigMode(normalizeComboConfigMode(settings?.comboConfigMode)))
.catch(() => setComboConfigMode("guided"));
+ fetch("/api/settings/compression")
+ .then((r) => (r.ok ? r.json() : null))
+ .then((settings) => setPromptCompressionEnabled(settings?.enabled === true))
+ .catch(() => setPromptCompressionEnabled(false));
fetch("/api/settings/proxy")
.then((r) => (r.ok ? r.json() : null))
.then((c) => setProxyConfig(c))
@@ -1128,6 +1133,7 @@ export default function CombosPage() {
<ComboCard
combo={combo}
metrics={metrics[combo.name]}
+ compressionEnabled={promptCompressionEnabled}
providerNodes={providerNodes}
copied={copied}
onCopy={copy}
@@ -1499,6 +1505,7 @@ function ComboReadinessPanel({ checks, blockers, showDescription = true }) {
function ComboCard({
combo,
metrics,
+ compressionEnabled,
copied,
onCopy,
onEdit,
@@ -1524,10 +1531,20 @@ function ComboCard({
const tc = useTranslations("common");
const emailsVisible = useEmailPrivacyStore((s) => s.emailsVisible);
const strategyDescription = getStrategyDescription(t, strategy);
- const initialCompressionMode = combo?.config?.compressionMode || combo.compressionOverride || "";
+ const hasRuntimeConfig = combo?.config && typeof combo.config === "object";
+ const initialCompressionMode =
+ typeof combo?.config?.compressionMode === "string"
+ ? combo.config.compressionMode
+ : hasRuntimeConfig
+ ? ""
+ : combo.compressionOverride || "";
const [compressionOverride, setCompressionOverride] = useState(initialCompressionMode);
const [isSavingCompression, setIsSavingCompression] = useState(false);
+ useEffect(() => {
+ setCompressionOverride(initialCompressionMode);
+ }, [initialCompressionMode]);
+
const handleCompressionOverrideChange = async (value) => {
setCompressionOverride(value);
setIsSavingCompression(true);
@@ -1687,20 +1704,22 @@ function ComboCard({
</span>
</div>
<div className="flex items-center gap-1.5 transition-opacity">
- <select
- value={compressionOverride}
- onChange={(e) => handleCompressionOverrideChange(e.target.value)}
- disabled={isSavingCompression}
- className="text-xs py-1 px-2 rounded border border-black/10 dark:border-white/10 bg-white dark:bg-bg-main text-text-main focus:border-primary focus:outline-none transition-colors disabled:opacity-50 max-w-[130px] md:max-w-none"
- title="Compression Override"
- >
- <option value="">Default</option>
- <option value="off">Off</option>
- <option value="lite">Lite</option>
- <option value="standard">Standard</option>
- <option value="aggressive">Aggressive</option>
- <option value="ultra">Ultra</option>
- </select>
+ {compressionEnabled && (
+ <select
+ value={compressionOverride}
+ onChange={(e) => handleCompressionOverrideChange(e.target.value)}
+ disabled={isSavingCompression}
+ className="text-xs py-1 px-2 rounded border border-black/10 dark:border-white/10 bg-white dark:bg-bg-main text-text-main focus:border-primary focus:outline-none transition-colors disabled:opacity-50 max-w-[130px] md:max-w-none"
+ title="Compression Override"
+ >
+ <option value="">Default</option>
+ <option value="off">Off</option>
+ <option value="lite">Lite</option>
+ <option value="standard">Standard</option>
+ <option value="aggressive">Aggressive</option>
+ <option value="ultra">Ultra</option>
+ </select>
+ )}
<button
onClick={onTest}
disabled={testing}
diff --git a/tests/integration/chatcore-compression-integration.test.ts b/tests/integration/chatcore-compression-integration.test.ts
index df3101b79a..00cd119beb 100644
--- a/tests/integration/chatcore-compression-integration.test.ts
+++ b/tests/integration/chatcore-compression-integration.test.ts
@@ -47,6 +47,12 @@ test("chatCore integration: compressContext called proactively when context exce
const provider = "openai";
const model = "gpt-4";
+ await compressionDb.updateCompressionSettings({
+ enabled: true,
+ defaultMode: "off",
+ autoTriggerTokens: 0,
+ });
+
// Create multiple messages with history that can be compressed
// Use the same pattern as test 3 which successfully tests compression
const body = {
@@ -120,9 +126,113 @@ test("chatCore integration: compressContext called proactively when context exce
}
});
+test("chatCore integration: disabled prompt compression leaves combo override requests unchanged", async () => {
+ const provider = "openai";
+ const model = "gpt-4";
+ const originalContextLength = process.env.CONTEXT_LENGTH_OPENAI;
+ process.env.CONTEXT_LENGTH_OPENAI = "8192";
+
+ await compressionDb.updateCompressionSettings({
+ enabled: false,
+ defaultMode: "off",
+ autoTriggerTokens: 1,
+ comboOverrides: { "disabled-compression-combo": "lite" },
+ });
+
+ const connection = await providersDb.createProviderConnection({
+ provider,
+ apiKey: "test-key",
+ isActive: true,
+ });
+
+ await combosDb.createCombo({
+ name: "disabled-compression-combo",
+ strategy: "priority",
+ models: [
+ {
+ kind: "model",
+ model: `${provider}/${model}`,
+ connectionId: connection.id,
+ },
+ ],
+ config: {
+ compressionMode: "lite",
+ },
+ });
+
+ const body = {
+ model: "combo/disabled-compression-combo",
+ stream: false,
+ messages: [
+ { role: "system", content: "You are helpful." },
+ { role: "user", content: `${"Keep spacing.\n\n\n".repeat(2000)}First long turn.` },
+ { role: "assistant", content: "Response 1" },
+ { role: "user", content: `${"Keep spacing.\n\n\n".repeat(2000)}Second long turn.` },
+ { role: "assistant", content: "Response 2" },
+ { role: "user", content: `${"Keep spacing.\n\n\n".repeat(2000)}Final question.` },
+ ],
+ };
+ const contextLimit = getTokenLimit(provider, model);
+ const proactiveThreshold = Math.floor(contextLimit * 0.7);
+ assert.ok(
+ estimateTokens(JSON.stringify(body.messages)) > proactiveThreshold,
+ "Test body should exceed proactive compression threshold"
+ );
+
+ let capturedBody: any = null;
+ globalThis.fetch = async (_url: string | URL | Request, init?: RequestInit) => {
+ if (init?.body) {
+ capturedBody = JSON.parse(init.body as string);
+ }
+ return new Response(
+ JSON.stringify({
+ choices: [{ message: { role: "assistant", content: "test" } }],
+ usage: { prompt_tokens: 10, completion_tokens: 5, total_tokens: 15 },
+ }),
+ {
+ status: 200,
+ headers: { "content-type": "application/json" },
+ }
+ );
+ };
+
+ try {
+ const result = await handleChatCore({
+ body,
+ modelInfo: { provider, model },
+ credentials: { apiKey: "test-key" },
+ log: { debug: () => {}, info: () => {}, warn: () => {}, error: () => {} },
+ clientRawRequest: { endpoint: "/v1/chat/completions", headers: new Map() },
+ connectionId: connection.id,
+ isCombo: true,
+ comboName: "disabled-compression-combo",
+ });
+
+ assert.ok(result.success, "Request should succeed");
+ assert.ok(capturedBody, "Fetch should have been called");
+ assert.deepEqual(capturedBody.messages, body.messages);
+
+ const summary = compressionAnalyticsDb.getCompressionAnalyticsSummary();
+ assert.equal(summary.totalRequests, 0, "Disabled compression should not record analytics");
+ } finally {
+ globalThis.fetch = originalFetch;
+ if (originalContextLength === undefined) {
+ delete process.env.CONTEXT_LENGTH_OPENAI;
+ } else {
+ process.env.CONTEXT_LENGTH_OPENAI = originalContextLength;
+ }
+ }
+});
+
test("chatCore integration: compressContext NOT called when context is below 85% threshold", async () => {
const provider = "openai";
const model = "gpt-4";
+
+ await compressionDb.updateCompressionSettings({
+ enabled: true,
+ defaultMode: "off",
+ autoTriggerTokens: 0,
+ });
const contextLimit = getTokenLimit(provider, model);
const threshold = Math.floor(contextLimit * 0.85);
@@ -199,6 +309,12 @@ test("chatCore integration: compression preserves message structure", async () =
const provider = "openai";
const model = "gpt-4";
+ await compressionDb.updateCompressionSettings({
+ enabled: true,
+ defaultMode: "off",
+ autoTriggerTokens: 0,
+ });
+
const body = {
model,
stream: false,
@@ -343,6 +459,12 @@ test("chatCore integration: combo requests run proactive compression before Kiro
const provider = "kiro";
const model = "claude-sonnet-4.5";
+ await compressionDb.updateCompressionSettings({
+ enabled: true,
+ defaultMode: "off",
+ autoTriggerTokens: 0,
+ });
+
const connection = await providersDb.createProviderConnection({
provider,
apiKey: "test-key",
diff --git a/tests/unit/compression/strategySelector.test.ts b/tests/unit/compression/strategySelector.test.ts
index b55af6d56a..d9c6a3dc5a 100644
--- a/tests/unit/compression/strategySelector.test.ts
+++ b/tests/unit/compression/strategySelector.test.ts
@@ -65,6 +65,17 @@ describe("getEffectiveMode", () => {
assert.equal(getEffectiveMode(config, null, 100), "off");
});
+ it("keeps disabled config off despite combo override and auto-trigger", () => {
+ const config = {
+ ...baseConfig,
+ enabled: false,
+ autoTriggerTokens: 100,
+ comboOverrides: { "my-combo": "lite" as const },
+ };
+
+ assert.equal(getEffectiveMode(config, "my-combo", 500), "off");
+ });
+
it("returns default mode when no overrides", () => {
assert.equal(getEffectiveMode(baseConfig, null, 100), "lite");
});