[chaitanyagiri]

What & why

Evening run, day 45 (11 Nov 2026): three new pages and one fix. Nothing launched after the 13:00 IST check that needed a page of its own, so the slots went to the next two High competitor rows and the top backlog row.

Merge order: #839, then #840, then #841, then this PR. This branch starts from #841's head (seo-engine/day-45-am), with base main.

• New: reason-machines-alternatives (1,148 words, 3 scenes and a title card: pulley, windmill, lantern). A competitor post (High row): Devin, Munder Difflin, OpenAI Codex cloud, Claude Code on the web and Cursor Cloud Agents.
• New: maxxwell-alternatives (1,165 words, 3 scenes and a title card: gears, pinball, spinningtop). A competitor post (High row): Orca, Munder Difflin, Paradev, AionUi and Agent Deck.
• New: what-is-nvidia-openshell (1,130 words, 3 scenes and a title card: terrarium, safe, airlock). An explainer for NVIDIA's agent sandbox, which shipped v0.1.8 on 10 Nov (backlog row b069).
• Fix: grok-bot-alternatives trimmed from 1,713 to 1,163 words, the top item in the refresh queue. It was the last part of the fix started in seo-engine: day 44 AM, Agent Tools Today 10 Nov, AionUi alternatives, Claude Code privacy #838. It adds no new facts; the old facts keep their 30 Sep dates.
• Links: one sentence each in orchestrator-inc-vs-munder-difflin (to Reason Machines alternatives), conductor-alternatives (to Maxxwell alternatives) and pi-vs-claude-code (to the OpenShell page).

Type of change

• Bug fix
• New feature
• Refactor / cleanup
• Docs
• Build / CI

Checks

• gate.mjs (full, with media and motion): passed for all three new pages, grok-bot-alternatives and the three linking posts.
• Fact review (two reviewers, sources re-fetched live on 11 Nov):
  • reason-machines-alternatives: FAIL once on 3 points, all fixed with the reviewer's text: Reason publishes more open source code than the network engine (Aspen, OpenAra, reason-plugins), it also runs on the web, and a cloud session "counts toward your plan's usage limits".
  • maxxwell-alternatives: FAIL once on 2 points, fixed: Claude Squad has had no release or commit since 20 Aug 2026, so it moved to "What did we leave out" and Agent Deck (MIT, v1.16.38 on 7 Nov) took its place, with its facts read from its README; the Orca quote now comes from the site's FAQ text.
  • what-is-nvidia-openshell: FAIL once on 5 points, all fixed with the reviewer's text: a release note quote had lost its qualifier, the upgrade note count, one unsourced "very safe", an FAQ line on model costs, and one word in an alt text.
  • grok-bot-alternatives trim: FAIL once on 6 points where the trim had dropped a qualifier or an attribution; all six put back with the reviewer's text.
  • Linking sentences: PASS.
• Scenes: each builder checked the frames at 25%, 50% and 75% and the stills; I checked a sheet of all twelve stills and the rendered pages. No Math.random in any motion.js.
• npm run build: passed (475 files written).

Notes

• OpenSEO is still signed out, so there was no Search Console pull, learn step or demand check this run.
• Apple's "Harbor" app was a lead from Hacker News; 9to5Mac (10 Nov) says it is in a private preview under the Claris brand, and Claris has no page for it, so no page here.
• Backlog row b077 (the Claude Code extension malware) is left for now: its facts are disputed and it is 13 days old.

Evidence

Images are on branch evidence/day-45-pm.

Before

The blog hub, the Grok Bot alternatives page and the three linking posts on #841's head.

After

The blog hub, the three new pages, the trimmed page, the three title cards and one scene from each new page.

🤖 Generated with Claude Code