Dev argues subagent fleets are pointless with Claude Opus 5.5 — one-on-one prompting wins
ryunuck · x · 2026-10-29
The author argues that with Opus 5.5, using subagents is useless because the model finally steers well from your own prompts and no longer talks weird, making deeply engaged one-on-one vibe coding far more effective than dispatching large agent fleets. Fleets dilute your personal "alpha" and pile up code debt, yielding the same results in the same time with less of your own qualities. Counterintuitively, someone ending the week at 30% usage while staying mentally present achieves more than a tokenmaxxing fleet dispatcher. The highest-value change should always be surgical and targeted — an elite sniper beats squadron sweeps.
More from coding & agent
- Google's PLE benchmark: best agent scores just 50.6% on real physical engineering tasks — rohanpaul_ai · 2026-10-29
- GitHub's zeeg: worktrees are the poor man's sandbox for coding agents — rsasaki0109 · 2026-10-29
- Building with coding agents and middle management are wildly different skills — corbtt · 2026-10-29
- Replit's ViBench 2.0: top models pass only 49-65% of long-horizon builds — amasad · 2026-10-29
- Interactive 3D biology lesson built with Opus 5.5 and Three.js follows one breath — techartist_ · 2026-10-29
- Replit's ViBench 2.0: best models pass only 49–65% of builds, GPT-6.1 Sol leads — amasad · 2026-10-29