@jiggai/recipes 0.4.62 → 0.4.63

package/openclaw.plugin.json

@@ -2,7 +2,7 @@
   "id": "recipes",
   "name": "Recipes",
   "description": "Markdown recipes that scaffold agents and teams (workspace-local).",
-  "version": "0.4.62",
+  "version": "0.4.63",
   "configSchema": {
     "type": "object",
     "additionalProperties": false,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jiggai/recipes",
-  "version": "0.4.62",
+  "version": "0.4.63",
   "description": "ClawRecipes plugin for OpenClaw (markdown recipes -> scaffold agents/teams)",
   "main": "index.ts",
   "type": "commonjs",

package/src/toolsInvoke.ts

@@ -47,7 +47,15 @@ function parseToolsInvokeError(json: ToolsInvokeResponse, status: number): strin
 
 async function doSingleToolsInvoke<T>(url: string, token: string, req: ToolsInvokeRequest): Promise<T> {
   const ac = new AbortController();
-  const t = setTimeout(() => ac.abort(), TOOLS_INVOKE_TIMEOUT_MS);
+  // Honor per-call timeoutMs (e.g. LLM nodes often pass 300000-900000 ms).
+  // Add a 30s HTTP/gateway overhead buffer so the LLM has the full node-
+  // configured window to produce its response before the fetch aborts.
+  // Fall back to the module default when no timeout was passed.
+  const argsTimeout = typeof (req.args as Record<string, unknown> | undefined)?.['timeoutMs'] === 'number'
+    ? (req.args as { timeoutMs: number }).timeoutMs
+    : 0;
+  const fetchTimeoutMs = argsTimeout > 0 ? argsTimeout + 30_000 : TOOLS_INVOKE_TIMEOUT_MS;
+  const t = setTimeout(() => ac.abort(), fetchTimeoutMs);
   const res = await fetch(url, {
     method: "POST",
     signal: ac.signal,