Reviews

Factory AI feature areas reviewed after enough real usage to have an earned opinion — not launch-day impressions from the changelog. The natural candidates are the big surfaces (Factory App, Droid CLI, Missions, Software Factory, remote delegations); smaller features earn pages when there's something to say.

Structure per review:

1. What it claims to be
2. What I ran it against
3. Where it broke
4. What stuck around after the test
5. Verdict

Rules:

• Ground every claim in observed behavior from my own usage notes. No usage evidence, no review — write the stub and leave ## Verdict as Pending.
• Never fabricate benchmark numbers or experience.
• When Factory ships the fix for something a review dinged, the review gets updated, not quietly buried. Updates get dated.