@askalf/dario 4.8.139 → 4.8.140

package/dist/cc-template.d.ts

@@ -432,6 +432,25 @@ export declare function resolveEffort(flag: EffortValue | undefined, clientBody:
  * — never a broken request.
  */
 export declare function supportsAdaptiveThinking(modelId: string): boolean;
+/**
+ * Anthropic prompt-cache control. `ttl` is the cache lifetime: `5m` (the
+ * upstream default when omitted) or `1h`. Real CC sends `1h` on its system
+ * blocks (see live-fingerprint.ts extractTemplate), so dario mirrors it.
+ */
+export type CacheControl = {
+    type: 'ephemeral';
+    ttl?: '5m' | '1h';
+};
+/**
+ * The cache-control dario stamps on every breakpoint (2 system + tools +
+ * conversation). `ttl: '1h'` mirrors real CC — dario shipped the 5-min
+ * ephemeral default, so its prefix expired 12x sooner than CC's and any
+ * interactive turn past the 5-min window re-created the whole prefix
+ * (system + all tools) at cache-creation cost, draining the Max window far
+ * faster than direct CC (root cause of dario#678). Single source of truth so
+ * the emitted TTL can't silently drift back to 5m.
+ */
+export declare const CC_CACHE_CONTROL: CacheControl;
 /**
  * Place CC-style prompt-cache breakpoints on the tools array and the
  * conversation. The system prompt is already cached at build time (2 system
@@ -446,12 +465,8 @@ export declare function supportsAdaptiveThinking(modelId: string): boolean;
  * sails through. CC genuinely caches tools + conversation, so NOT caching them
  * was itself a wire divergence from CC. Exported for unit testing.
  */
-export declare function applyCcPromptCaching(ccRequest: Record<string, unknown>, cacheControl: {
-    type: 'ephemeral';
-}): void;
-export declare function buildCCRequest(clientBody: Record<string, unknown>, billingTag: string, cacheControl: {
-    type: 'ephemeral';
-}, identity: {
+export declare function applyCcPromptCaching(ccRequest: Record<string, unknown>, cacheControl: CacheControl): void;
+export declare function buildCCRequest(clientBody: Record<string, unknown>, billingTag: string, cacheControl: CacheControl, identity: {
     deviceId: string;
     accountUuid: string;
     sessionId: string;

package/dist/cc-template.js

@@ -1235,6 +1235,16 @@ export function supportsAdaptiveThinking(modelId) {
         return true;
     return false;
 }
+/**
+ * The cache-control dario stamps on every breakpoint (2 system + tools +
+ * conversation). `ttl: '1h'` mirrors real CC — dario shipped the 5-min
+ * ephemeral default, so its prefix expired 12x sooner than CC's and any
+ * interactive turn past the 5-min window re-created the whole prefix
+ * (system + all tools) at cache-creation cost, draining the Max window far
+ * faster than direct CC (root cause of dario#678). Single source of truth so
+ * the emitted TTL can't silently drift back to 5m.
+ */
+export const CC_CACHE_CONTROL = { type: 'ephemeral', ttl: '1h' };
 /**
  * Place CC-style prompt-cache breakpoints on the tools array and the
  * conversation. The system prompt is already cached at build time (2 system

package/dist/doctor.js

@@ -334,7 +334,7 @@ export async function runChecks(opts = {}) {
     // non-passthrough request and dominate the input-token cost on small
     // turns. Anthropic caches them after the first hit (cache_creation
     // tokens on call 1, then cache_read on subsequent calls within the
-    // 5-min/1-hr TTL), but non-CC users routing heavy tooling get
+    // 1-hour TTL), but non-CC users routing heavy tooling get
     // surprised by the first-request charge. Surface the size up front
     // so they can plan.
     //
@@ -355,7 +355,7 @@ export async function runChecks(opts = {}) {
                 detail: `${promptChars.toLocaleString()} chars system prompt + ${toolCount} tool defs ` +
                     `(${toolChars.toLocaleString()} chars JSON-serialized) injected per non-passthrough ` +
                     `request. Cached after first hit; read-cost only on subsequent calls within ` +
-                    `the 5-min/1-hr TTL. Exact token count surfaces as cache_creation_input_tokens ` +
+                    `the 1-hour TTL. Exact token count surfaces as cache_creation_input_tokens ` +
                     `on the first response (or run \`dario doctor --usage\`).`,
             });
         }

package/dist/proxy.js

@@ -9,7 +9,7 @@ import { arch, platform } from 'node:process';
 import { getAccessToken, getStatus } from './oauth.js';
 import { buildHealthResponse, derivePoolStatus, shouldDiscloseHealthInternals } from './health-response.js';
 import { darioVersion } from './version.js';
-import { buildCCRequest, applyCcPromptCaching, parseEffortSuffix, reverseMapResponse, createStreamingReverseMapper, orderHeadersForOutbound, isMcpToolName, CC_TEMPLATE } from './cc-template.js';
+import { buildCCRequest, applyCcPromptCaching, parseEffortSuffix, reverseMapResponse, createStreamingReverseMapper, orderHeadersForOutbound, isMcpToolName, CC_TEMPLATE, CC_CACHE_CONTROL } from './cc-template.js';
 import { stampCch, hasCchSeed } from './cch.js';
 import { describeTemplate, detectDrift, checkCCCompat } from './live-fingerprint.js';
 import { AccountPool, computeStickyKey, parseRateLimits, modelFamily, isInAuthCooldown, authCooldownMs, reconcilePoolAccounts } from './pool.js';
@@ -2049,7 +2049,8 @@ export async function startProxy(opts = {}) {
                         // current Claude Code, which sends none. dario#528.
                         const cch = hasCchSeed(cliVersion) ? computeCch() : null;
                         const billingTag = buildBillingTag(cliVersion, cch);
-                        const CACHE_EPHEMERAL = { type: 'ephemeral' };
+                        // 1h cache TTL, matching real CC (see CC_CACHE_CONTROL / dario#678).
+                        const CACHE_EPHEMERAL = CC_CACHE_CONTROL;
                         // Session stickiness: rebind the pre-selected pool account to
                         // whatever the sticky-key resolver picks. If this is a new
                         // conversation the key binds to the current best account

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@askalf/dario",
-  "version": "4.8.139",
+  "version": "4.8.140",
   "description": "Use your Claude Pro/Max subscription in any tool — Cursor, Cline, Aider, the Agent SDK, your scripts — at subscription pricing, not per-token API bills. One local Anthropic + OpenAI-compatible endpoint.",
   "type": "module",
   "bin": {