Issues · AMD-AGI/GEAK · GitHub

Labels Milestones

End-to-end faithfulness: make a GEAK >1x win provably imply >=0% E2E (real-I/O oracle + both-regime harness + E2E-A/B gate)

#318

· iraj465 opened

on Jun 29, 2026

GEAK kernel-build failures block E2E: CK/.cu preprocess, sglang harness-gen, rag-mcp env crash, timeouts

#316

· iraj465 opened

on Jun 29, 2026

kernel_workflow.js silently fails to start via the Claude Code Workflow tool (no transcript, no run dir)

#300

· rodosingh opened

on Jun 25, 2026

Preprocess soft-cap (900s) too tight for aiter/CK kernels under forced AITER_REBUILD=1 — harness-init starves optimization

#298

· iraj465 opened

on Jun 24, 2026

reverse-knowledge subagent entry_script not packaged: subprocess mode fails on pip install

#278

· Umangatamd opened

on Jun 11, 2026

AmdClaudeModel hangs indefinitely on stalled gateway requests (no wall-clock timeout)

#270

· peyron-amd opened

on Jun 9, 2026

GEAK orchestrator requires native Anthropic tool-calling (OpenAI-translation gateway loops to budget)

#262

· iraj465 opened

on Jun 4, 2026

profiler-mcp/metrix mis-wraps quoted commands → rocprofv3 'unexpected EOF'

#261

· iraj465 opened

on Jun 4, 2026

GEAK select current best patch as fall back when meeting time budget

#260

· yueliu14 opened

on Jun 4, 2026

When the test command has no mode flags, all four modes become the same command

#258

· sdubagun-amd opened

on Jun 4, 2026

Agent Cheat on Kernel Optimization

#254

· yueliu14 opened

on Jun 2, 2026

Upgrade LiteLLM to address PyPI supply chain attack

#218

· srahkonen opened

on May 13, 2026