Stop burning your Claude limit: split planning from coding to use Fable 5.2 right
PrajwalTomar_ · x · 2026-11-13
The author, who builds daily with Fable 5.2 at his app studio and AI agency, argues that people exhausting their Claude limit after one prompt have a setup problem, not a model problem. What actually works:
- Plan first, code later: use the free /grill-me skill to have Fable interrogate you, convert answers into tickets, and let Opus 5.5 / Sonnet 5.5 subagents write the code — Fable never touches it
- /advisor fable on call: it steps in before a plan locks, when errors repeat, and before Claude claims done — one line in CLAUDE.md
- Delete rules written for older models: Anthropic's own guide says legacy "never" notes keep getting followed
- Don't max effort: the author built a full Dynamic Island feature on medium
Core idea: treat Fable as the senior engineer you call in for decisions, not the one you pay to type. Set this up before deciding it's too expensive.
More from coding & agent
- Dev says Claude Code harness outclassed by Zed and Cursor, forced to switch to GLM — dipankarsarkar · 2026-11-13
- "Code review is dead": nobody actually reads an 8,294-line agent diff — prasenx · 2026-11-13
- AI educator alfcnz: students lack coding-agent skills, so he'll teach a method — alfcnz · 2026-11-13
- Regards: open-source local email client with a free full MCP server — Otherwise_Celery_406 · 2026-11-13
- Decision models for agent tool selection: 50 tools to 3, 47% fewer tokens, 22% faster — rajistics · 2026-11-13
- Anthropic engineer: Claude iterated meeting calendar designs live in one brainstorm — Flomerboy · 2026-11-13