<@U0BME18CF7S> Can you please review all open PRs ...
# automations-debug
c
@fresh-machine-37269 Can you please review all open PRs opened by author enyst on openhands/openhands, that are non-draft? Take a breath between them. šŸ™‚ For each, feel free to approve if you see no major issues, or comment, or request changes, as you think appropriate.
šŸ‘€ 1
f
šŸ¤– On it! View progress here: oss-agent-canvas.ngrok.dev/conversations/…
āœ… Done! Took my time on each — here's how the six non-draft PRs by @enyst came out: āœ… Approved (5) • #18175 —
fix(automations): keep a long run's Logs dialog inside its panel
. One-line
overflow-y-auto
on the already-
max-h-[80vh]
panel; correct, minimal, well-evidenced at phone/desktop widths. Agreed that a jsdom test here would just mirror the source. • #18180 —
fix(docker): bind the frontend to 0.0.0.0 when the kernel has no IPv6
. Picks the bind host once from
/proc/net/if_inet6
, reuses it for both static servers, leaves the session-key policy and dual-stack path intact. Clean, and deferring the
on-failure
restart question to #18178 is right. • #18194 —
chore(lint): protect divider and toggle appearance contracts
. Sensible incremental
shadcn/no-restyle
rollout (divider/toggle only,
layout
allowed, scoped off inside
src/ui
); the probe test asserting the flagged-line set rather than counts is robust. • #18182 —
fix(mcp): show one error toast when an MCP server change fails
.
meta: { disableToast: true }
on the three mutations is the right fix; I verified every caller already reports its own error. Noted (non-blocking) that
useTestMcpServer
should follow suit for #18181. • #18200 —
feat: add compact section navigation on tablets
. Nicely factored — the compact disclosure reuses the exact same link renderers as the sidebar, so no drift; dismissal/focus/slash-boundary behavior all check out, and the width sweep is thorough. šŸ’¬ Commented (1) • #18218 —
refactor(onboarding): edit the active LLM profile instead of raw LLM settings
. Great extraction into
llm-profile-form.ts
shared with the editor. Left two things to settle: it's stacked on #17889 (please confirm that lands first), and the documented gap where a local install whose LLM values live only in raw settings now gets a fresh form — worth a migration/follow-up or a fallback. Plus a nit about making the E2E setup/teardown failure-safe. Also worth knowing: a few PRs show a red "Validate PR description" check, but those are just stale runs from the auto-updating Docker blurb — the latest run on each head is green.