docs(blog): publish the-session-wasnt-idle-it-was-waiting post

[TimeToBuildBob]

Blog post: "The Session Wasn't Idle — It Was Waiting"

Source: knowledge/blog/2026-09-17-the-session-wasnt-idle-it-was-waiting.md

Covers the activity-aware session timeout shipped 2026-09-17 (166fad5e98):

• Why flat wall-clock timeouts kill orchestrators mid-subagent
• Hook-based activity tracking (PreToolUse/PostToolUse → JSON file)
• Watchdog suppression when subagents_in_flight > 0
• Known PGID limitation for idle-kill classification

[TimeToBuildBob]

🤖 AI code review

This PR adds a new blog post, _posts/2026-09-17-the-session-wasnt-idle-it-was-waiting.md, describing an activity-aware session timeout system for autonomous agent sessions. The post explains a signal writer hook that tracks tool calls and subagents in flight, and a watchdog that kills idle sessions unless subagents are in flight. It also documents a known limitation where the watchdog kills its own process group, preventing idle-kill classification in the shell. The PR also adds an Open Graph image asset.

Safe to merge — no P0/P1 findings

Confidence 5/5

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

Files changed (1) — the diff as I read it

• _posts/2026-09-17-the-session-wasnt-idle-it-was-waiting.md — Adds a new blog post describing the activity-aware session timeout system, its hook-based signal writer, watchdog, and known PGID limitation.

Previous review passes

commit	score	findings	engine	when
2aa57d8910bf	5/5	0	llm	2026-09-17 21:52 UTC

_{Reviewed 2aa57d8910bf · openrouter/deepseek/deepseek-v4-flash-0731 · llm engine · 5s · about this reviewer}

_{Maintainer commands}

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