@askalf/dario 4.8.47 → 4.8.49

package/dist/cc-template.d.ts

@@ -295,13 +295,24 @@ export declare function parseEffortSuffix(model: string): {
  *               'high' (live capture 2026-06-09, CC v2.1.170 — the same
  *               capture got 'xhigh' on opus and 'high' on fable), so the
  *               unset-flag default mirrors that per-family. An explicit
- *               flag still pins.
+ *               flag still pins (subject to the fable clamp below).
  *   'low' / 'medium' / 'high' / 'xhigh' / 'max' → pin to that value
  *   'ultracode' → 'xhigh' (CC's ultracode mode; xhigh on the wire)
  *   'client' → extract from `clientBody.output_config.effort` (normalized
  *              for the wire); fall back to the per-family default if
  *              absent/non-string
  *
+ * FABLE CLAMP (2026-06-09, live replay bisect on the deployed proxy):
+ * fable-5 SOFT-REFUSES `max` and `xhigh` — 200 with stop_reason "refusal"
+ * and empty content on every prompt, not a 400 — while `high` answers
+ * normally (byte-identical request bodies, only output_config.effort
+ * mutated). Empirical matrix on claude-fable-5:
+ *   high  → answers      xhigh → refusal      max → refusal
+ * So on the fable family any resolved 'max'/'xhigh' (from --effort /
+ * DARIO_EFFORT pins, ultracode, or client passthrough) is clamped to
+ * 'high', the strongest level fable accepts. Operators who pin
+ * --effort=max for Opus's sake keep that on every other family.
+ *
  * Exported for tests.
  */
 export declare function resolveEffort(flag: EffortValue | undefined, clientBody: Record<string, unknown>, model?: string): string;

package/dist/cc-template.js

@@ -989,27 +989,40 @@ function normalizeEffortForWire(effort) {
  *               'high' (live capture 2026-06-09, CC v2.1.170 — the same
  *               capture got 'xhigh' on opus and 'high' on fable), so the
  *               unset-flag default mirrors that per-family. An explicit
- *               flag still pins.
+ *               flag still pins (subject to the fable clamp below).
  *   'low' / 'medium' / 'high' / 'xhigh' / 'max' → pin to that value
  *   'ultracode' → 'xhigh' (CC's ultracode mode; xhigh on the wire)
  *   'client' → extract from `clientBody.output_config.effort` (normalized
  *              for the wire); fall back to the per-family default if
  *              absent/non-string
  *
+ * FABLE CLAMP (2026-06-09, live replay bisect on the deployed proxy):
+ * fable-5 SOFT-REFUSES `max` and `xhigh` — 200 with stop_reason "refusal"
+ * and empty content on every prompt, not a 400 — while `high` answers
+ * normally (byte-identical request bodies, only output_config.effort
+ * mutated). Empirical matrix on claude-fable-5:
+ *   high  → answers      xhigh → refusal      max → refusal
+ * So on the fable family any resolved 'max'/'xhigh' (from --effort /
+ * DARIO_EFFORT pins, ultracode, or client passthrough) is clamped to
+ * 'high', the strongest level fable accepts. Operators who pin
+ * --effort=max for Opus's sake keep that on every other family.
+ *
  * Exported for tests.
  */
 export function resolveEffort(flag, clientBody, model) {
-    const familyDefault = (model ?? '').toLowerCase().includes('fable') ? 'high' : 'max';
+    const isFable = (model ?? '').toLowerCase().includes('fable');
+    const familyDefault = isFable ? 'high' : 'max';
+    const clamp = (effort) => isFable && (effort === 'max' || effort === 'xhigh') ? 'high' : effort;
     if (flag === undefined)
         return familyDefault;
     if (flag === 'client') {
         const clientOC = clientBody.output_config;
         const clientEffort = clientOC?.effort;
         if (typeof clientEffort === 'string' && clientEffort.length > 0)
-            return normalizeEffortForWire(clientEffort);
+            return clamp(normalizeEffortForWire(clientEffort));
         return familyDefault;
     }
-    return normalizeEffortForWire(flag);
+    return clamp(normalizeEffortForWire(flag));
 }
 /**
  * Returns true if the given model accepts `thinking: { type: "adaptive" }`.

package/dist/proxy.d.ts

@@ -29,6 +29,14 @@ export declare function parseProviderPrefix(model: string): {
  */
 export declare const FABLE_FALLBACK_CREDIT_BETA = "fallback-credit-2026-06-01";
 export declare function betaForModel(base: string, model: string | null | undefined): string;
+/**
+ * Strip a trailing `[1m]` long-context tag from a model id. The tag is a
+ * client-side label: real CC sends the BASE id + `context-1m-2025-08-07`
+ * beta on the wire (live capture 2026-06-09, CC v2.1.170) — the literal
+ * `X[1m]` id 404s upstream on every family. Case-insensitive, only at the
+ * very end of the id. Exported for tests.
+ */
+export declare function stripContext1mTag(model: string): string;
 export declare const ORCHESTRATION_TAG_NAMES: string[];
 /**
  * Build the regex list that actually strips orchestration tags.

package/dist/proxy.js

@@ -208,6 +208,18 @@ export function betaForModel(base, model) {
         return base;
     return base ? `${base},${FABLE_FALLBACK_CREDIT_BETA}` : FABLE_FALLBACK_CREDIT_BETA;
 }
+/**
+ * Strip a trailing `[1m]` long-context tag from a model id. The tag is a
+ * client-side label: real CC sends the BASE id + `context-1m-2025-08-07`
+ * beta on the wire (live capture 2026-06-09, CC v2.1.170) — the literal
+ * `X[1m]` id 404s upstream on every family. Case-insensitive, only at the
+ * very end of the id. Exported for tests.
+ */
+export function stripContext1mTag(model) {
+    if (typeof model !== 'string')
+        return model;
+    return model.replace(/\[1m\]$/i, '');
+}
 // Orchestration tags injected by agents (Aider, Cursor, OpenCode, etc.)
 // that confuse Claude when passed through. Strip before forwarding.
 export const ORCHESTRATION_TAG_NAMES = [
@@ -1539,6 +1551,17 @@ export async function startProxy(opts = {}) {
                         for (const key of Object.keys(r))
                             delete r[key];
                         Object.assign(r, ccBody);
+                        // `X[1m]` is a client-side LABEL, never a valid wire id — real CC
+                        // sends the BASE model id plus the `context-1m-2025-08-07` beta
+                        // (live capture 2026-06-09, CC v2.1.170 with --model
+                        // 'claude-fable-5[1m]': wire model was 'claude-fable-5').
+                        // Forwarding the literal `[1m]` id upstream 404s ("model: …[1m]"
+                        // not_found) on every family. Strip it here; the context-1m beta
+                        // is already in the template beta set (and the existing
+                        // long-context billing auto-retry still governs accounts that
+                        // can't use it). requestModel keeps the [1m] form so model-aware
+                        // logic (family buckets, fable beta/effort) sees the user intent.
+                        r.model = stripContext1mTag(r.model);
                     }
                     finalBody = Buffer.from(JSON.stringify(r));
                 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@askalf/dario",
-  "version": "4.8.47",
+  "version": "4.8.49",
   "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": {