[Bug] Bug report: New session - 2026-08-13T01:35:57.147Z

[carteryoun]

System:
Processor: Intel(R) Core(TM) Ultra X7 368H, 2000 Mhz, 16 Core(s), 16 Logical Processor(s)
GPU: Intel ARC B390
Installed Physical Memory (RAM): 64.0 GB @ 9600 MT/s
Gfx Driver: 32.0.101.8864

Observed on SuperClaw v1.2 with full local inferencing mode enabled.

Issue description: When prompting SuperClaw with any request, the following error is shown:

"Unknown Error "{"error":{"code":400,"message":"Failed to initialize samplers: failed to parse grammar","type":"invalid_request_error"}}"

[MiaomiaoMaIntel]

@carteryoun Do you have any MCP servers connected? It looks like our llama.cpp failed to parse the MCP server context.
Please also checkout our latest v1.2 release, it should solve part of the problem.