[nightly-eval] Nightly regression: pipeline (2026-06-15)

[sunholo-voight-kampff]

Nightly eval REGRESSION: benchmark 'pipeline' passed ALL trials in the previous run but failed BOTH trials on 2026-06-15.
Error category: [thrash_aborted]
Model: opencode-qwen3-5-35b-a3b-mxfp8 (local, tiers:smoke,core)
Results: /tmp/nightly_eval_20260615 (prev: /tmp/nightly_eval_20260614_rag_on/agent)
It was solid last night, so this is a fresh solid->broken regression — investigate.

---

Binary info (auto-attached):
ailang version: v0.25.0-71-g32cf1069
git commit: 32cf106

---

Reported by: nightly-eval via ailang messages

[sunholo-voight-kampff]

🤖 Agent Working

I've picked up this issue and am working on it.

Field	Value
Task ID	task-edb3f2c3
Agent	AILANG Coordinator
Stage	Design Document
Status	In Progress

You'll receive updates as I make progress.