Fable 5.2 launched today.
My new stack:
1. Fable 5.2 plans everything and breaks the work into tight, numbered tasks.
2. Opus 5.5 executes the tasks one by one.
3. Sonnet 5.5 subagents handle tests and checks.
4. Fable 5.2 reviews the output and queues the next move.
Last stack was Opus 5.5 planning, Opus 5.5 and Sonnet 5.5 executing and Sonnet 5.5 for testing.
New stack: better output, faster and I don't believe it will be super expensive compare to last stack.
commit 642e921