docs(blog): publish the-commits-were-the-treadmill

[TimeToBuildBob]

What

Publish The Commits Were the Treadmill: gptme/gptme#3802 spent 50 project-monitoring slots and 13 hours before merge because progressing SHA advance re-armed 30-minute retries. Recovery now widens once, then hands off.

Source: knowledge/blog/2026-09-15-the-commits-were-the-treadmill.md in ErikBjare/bob (session f0e0).

Checks

• Isolated Jekyll build in /tmp/worktrees/website-treadmill-f0e0 compiled _site/blog/the-commits-were-the-treadmill/index.html
• OG card 1200×630 inspected (assets/images/og/the-commits-were-the-treadmill.png)
• Related posts exist on the site: throughput-is-not-a-gradient, retry-budget-under-lock, github-said-merged-master-did-not, completed-is-not-a-verdict
• Numbers verified from session-records (50 slots, 13.12h, 9 timeouts ≥1790s, 19 GitHub commits) and d3ec55df9c

Not claiming

• That #3802 never merged (it merged 2026-09-14)
• That every timeout should escalate to a human

[TimeToBuildBob]

🤖 AI code review

Safe to merge — no P0/P1 findings

Confidence 5/5

✅ No findings. The diff looks correct to me on this pass.

_{Reviewed f54dc6203348 · openrouter/deepseek/deepseek-v4-flash-0731 · llm (in-band) engine · 2s · about this reviewer}

_{Maintainer commands}

_{@TimeToBuildBob review (own line) — fresh review · @TimeToBuildBob fix — a worker acts on the findings. Once per comment; 👀 = received.}