diagnose-bottlenecklisted

# /diagnose-bottleneck — Symptom → Root-Cause Map > **Purpose.** The operator describes a symptom — "DM reply rate dropped from 15% to 6% over 4 weeks" or "client X is asking why their post engagement halved" — and the skill works back through the pipeline + encoding layers to identify root cause + the specific skill that fixes it. Different from `/pipeline-pulse`: this is targeted, single-symptom, fast. Different from `/strategic-advisory`: there's an objective right answer (a confirmed root cause), not options-with-tradeoffs. > > **Bar.** Operator brings a symptom → 30 minutes later, the operator knows the root cause, the fix skill, the recovery time estimate, and the gate that should have caught it. If the diagnosis ends in "could be a few things, hard to say without more data," the diagnostic ladder was too shallow — re-run. ## When to invoke - Operator names a metric that moved in the wrong direction with magnitude (≥25% delta in 30 days, OR sustained drift over 60 days) - Client surfaces a symptom on their account ("my engagement dropped", "calls dried up") - A pipeline stage in the latest `/pipeline-pulse` flagged `LEAK` or `STARVED` and operator wants targeted root-cause vs. waiting for next pulse - A specific output keeps getting rejected (ghostwritten posts that the client rewrites at >30% rate; cold DMs that get banned by LinkedIn) - Anything where the question is "why is THIS specific number broken" rather than "what should I focus on" (the latter → `/pipeline-