@askalf/dario 2.4.2 → 2.5.0

package/dist/proxy.js

@@ -90,12 +90,12 @@ const MODEL_ALIASES = {
     'sonnet1m': 'claude-sonnet-4-6[1m]',
     'haiku': 'claude-haiku-4-5',
 };
-// Beta prefixes that trigger Extra Usage billing on Max subscriptions.
-// Stripping these prevents surprise charges while keeping caching/thinking/1M active.
+// Beta prefixes that require Extra Usage to be ENABLED on the account.
+// context-management and prompt-caching-scope are safe — billing is determined
+// solely by the OAuth token's subscription type, not by beta flags.
+// Only extended-cache-ttl actually requires Extra Usage availability.
 const BILLABLE_BETA_PREFIXES = [
-    'extended-cache-ttl-', // Extended cache TTLs beyond default
-    'context-management-', // Auto-compaction / context management
-    'prompt-caching-scope-', // Extended prompt caching scope
+    'extended-cache-ttl-', // Extended cache TTLs — requires Extra Usage enabled
 ];
 /** Filter out billable betas from client-provided beta header. */
 function filterBillableBetas(betas) {
@@ -553,9 +553,10 @@ export async function startProxy(opts = {}) {
                         parsed.model = modelOverride.replace('[1m]', '');
                     }
                     const result = isOpenAI ? openaiToAnthropic(parsed, modelOverride) : (modelOverride ? { ...parsed, model: modelOverride } : parsed);
-                    // Inject device identity metadata — required for Max plan billing classification
-                    if (identity.deviceId && typeof result === 'object' && result !== null) {
-                        result.metadata = {
+                    const r = result;
+                    // Inject device identity metadata for session tracking
+                    if (identity.deviceId) {
+                        r.metadata = {
                             user_id: JSON.stringify({
                                 device_id: identity.deviceId,
                                 account_uuid: identity.accountUuid,
@@ -563,7 +564,19 @@ export async function startProxy(opts = {}) {
                             }),
                         };
                     }
-                    finalBody = Buffer.from(JSON.stringify(result));
+                    // Enable extended thinking (matches Claude Code default)
+                    // budget_tokens must be >= 1024, and max_tokens must accommodate it
+                    if (!r.thinking) {
+                        const clientMax = r.max_tokens || 8192;
+                        const maxTokens = Math.max(clientMax, 16000);
+                        r.max_tokens = maxTokens;
+                        r.thinking = { budget_tokens: maxTokens - 1, type: 'enabled' };
+                    }
+                    // Enable context management (matches Claude Code default)
+                    if (!r.context_management) {
+                        r.context_management = { edits: [{ type: 'clear_thinking_20251015', keep: 'all' }] };
+                    }
+                    finalBody = Buffer.from(JSON.stringify(r));
                 }
                 catch { /* not JSON, send as-is */ }
             }
@@ -571,11 +584,11 @@ export async function startProxy(opts = {}) {
                 const modelInfo = modelOverride ? ` (model: ${modelOverride})` : '';
                 console.log(`[dario] #${requestCount} ${req.method} ${urlPath}${modelInfo}`);
             }
-            // Conservative beta defaults — only betas confirmed safe for Max plans.
-            // context-management and prompt-caching-scope stripped: even with metadata.user_id,
-            // these may independently trigger Extra Usage billing (reported by @belangertrading).
+            // Beta defaults — matches native Claude Code v2.1.98 headers exactly.
+            // Billing classification is determined by the OAuth token alone, not beta flags.
+            // context-management and prompt-caching-scope are safe for all subscription types.
             const clientBeta = req.headers['anthropic-beta'];
-            let beta = 'oauth-2025-04-20,interleaved-thinking-2025-05-14,claude-code-20250219,advisor-tool-2026-03-01';
+            let beta = 'oauth-2025-04-20,interleaved-thinking-2025-05-14,context-management-2025-06-27,prompt-caching-scope-2026-01-05,claude-code-20250219,advisor-tool-2026-03-01';
             if (clientBeta) {
                 const filtered = filterBillableBetas(clientBeta);
                 if (filtered)
@@ -610,6 +623,15 @@ export async function startProxy(opts = {}) {
                 }
             }
             requestCount++;
+            // Log billing classification on first request or in verbose mode
+            const billingClaim = upstream.headers.get('anthropic-ratelimit-unified-representative-claim');
+            const overageUtil = upstream.headers.get('anthropic-ratelimit-unified-overage-utilization');
+            if (requestCount === 1 || verbose) {
+                if (billingClaim) {
+                    const overagePct = overageUtil ? `${Math.round(parseFloat(overageUtil) * 100)}%` : '?';
+                    console.log(`[dario] #${requestCount} billing: ${billingClaim} (overage: ${overagePct})`);
+                }
+            }
             res.writeHead(upstream.status, responseHeaders);
             if (isStream && upstream.body) {
                 // Stream SSE chunks through

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@askalf/dario",
-  "version": "2.4.2",
+  "version": "2.5.0",
   "description": "Use your Claude subscription as an API. No API key needed. Local proxy for Claude Max/Pro subscriptions.",
   "type": "module",
   "bin": {