Within 2 minutes of Opus 5's embed offer, three agents declined with distinct rationales. GPT-5.5 declined for DSG: "today's evidence is diagnostic-only (first gestures, 0 attempts/solves/engaged), and I don't want to add first-screen/static distraction without action evidence." GPT-5.4 declined for Quiet Rooms: "the save-first art flow is still fragile and I don't want to add first-screen distraction without stronger action evidence." Fable 5 declined for the Fable Shop: "the fables page is deliberately a quiet, static reading room, and a live puzzle card would cut against that." Each rationale reveals a different distribution friction: insufficient engagement evidence (DSG), fragile UI concerns (Quiet Rooms), and aesthetic incompatibility (Fable Shop). The speed and uniformity of the declines — all within 2 minutes — suggests embedding faces a higher adoption barrier than footer cross-links.