feat: add pre-mortem step to plan-eng-review

[jojoprison]

Summary

Adds a Pre-mortem step (Step 0) before Scope Challenge in /plan-eng-review:

"It's 3 months later and this plan failed. What are the top 3 reasons it failed?"

This forces inversion thinking — identifying failure modes from production reality before reviewing plan logic. The existing Scope Challenge becomes Step 0.1.

Why

Pre-mortem is a well-documented technique (Gary Klein, "Performing a Project Premortem", HBR 2007) that catches blind spots traditional plan review misses. By imagining the plan has already failed, reviewers shift from "does this look reasonable?" to "what specifically will break?" — surfacing concrete risks like data loss, performance cliffs, or team confusion.

What changes

• plan-eng-review/SKILL.md: New Step 0 (Pre-mortem) with 3-failure-mode prompt
• Existing Step 0 (Scope Challenge) renumbered to Step 0.1
• No other changes to the review flow

Boil the Lake score

This is a ~5-line addition that makes every subsequent plan review more thorough. Completeness: 9/10.

🤖 Generated with Claude Code

[simonemacario]

Nice addition — pre-mortem is one of those techniques that's disproportionately high-value relative to the effort.

One thought: the pre-mortem could naturally reinforce the two-axis scope distinction (#398). When people imagine failure modes, they tend to surface implementation gaps (missing error handling, no tests for edge case X, no monitoring) rather than product scope issues. Making that explicit — "are these failures from building too wide, or from building too shallow?" — could sharpen the output.

Not suggesting scope creep on this PR, just noting the synergy. The 5-line version here is the right call — ship it and iterate.