A week with Claude Opus 5.5: six tips to save your credits
every · x · 2026-09-23
After testing Claude Opus 5.5 for a week, the Every team shared six practical tips:
- Set a budget and stopping point: it won't stop on its own and will eat your weekly limit; define what "done" looks like
- Ask for the deliverable first: given 10 minutes for a run of show, it built a data generator and handouts, then ran out of time
- Let it critique itself, then check: it caught issues the designer missed, but rejected 90% of its own work for untraceable reasons
- Give it notes: 5.5 now builds on feedback instead of arguing
- Move the lede up yourself: most readable prose they've measured, but still buries the point
- Point it at visual work: game UI, 3D, animations and prototypes were its best one-shot results
Related event: Six Practical Tips After a Week of Testing Claude Opus 5.5(2 posts)→
More from coding & agent
- After a Weekend with Jev, Data Engineers Say It Could Rewrite the Workflow — hackgoofer · 2026-09-23
- Claude Design vs Figma MCP: Interactive HTML Prototypes, No Figma Dependency — dotey · 2026-09-23
- Reef adds GEPA prompt optimization and Meta-Harness evolution for self-improving agents — pliang279 · 2026-09-23
- 60s local AI video on a 5060 Ti: 30-clip pipeline catches agent tool-drift with a Python audit layer — Wonderful_Sample6291 · 2026-09-23
- Prompts will be the new App Store content: Kneen lays out Idea > Spec > Applets > App dev flow — jasonkneen · 2026-09-23
- Anthropic's official prompt stops Opus in Claude Code from pausing mid-task to report — omarsar0 · 2026-09-23