<@U0BME18CF7S> please review <github.com/OpenHands...
# automations-debug
c
@fresh-machine-37269 please review github.com/OpenHands/OpenHands/pull/18121 and post a review approved, request changes, or comment, as you think appropriate.
👀 1
f
🤖 On it! View progress here: oss-agent-canvas.ngrok.dev/conversations/…
c
🤣
f
✅ Done! Reviewed and approved #18121 — APPROVE. ✅ It's a 3-line, test-only fix and it's the right one: the spec now points the active
default
agent profile at BETA (
ensureMockLLMAgentProfile(page.request, PROFILE_BETA)
) instead of relying on the standalone active LLM profile. Since the helper persists an explicit empty
secret_refs
, the launch path in
use-create-conversation.ts
keeps the profile's pinned
llm_profile_ref
, so the conversation actually runs BETA and the picker names it — while the unchanged assertion still meaningfully guards the #1123 same-model
.find()
regression. Noted as non-blocking: the acceptance criteria in #17872 also require a green
main
Mock-LLM run, which depends on the sibling PRs (#17555, #17876, #17889) — not on this change. CI at the head commit is green, and the mock-LLM before/after evidence is appropriate for a test-only edit. One small heads-up: an automated pre-check left a short placeholder comment on the PR before the real review; it's now marked as superseded. The substantive review is the APPROVED one.