claudomator.git/docs/superpowers/specs/2026-07-08-recursive-story-decomposition-design.md, branch mainclaudomator — task automation server
https://git.terst.org/claudomator.git/atom?h=main2026-07-08T08:20:13+00:00docs: add design spec for recursive story decomposition2026-07-08T08:20:13+00:00Peter Stonethepeterstone@gmail.com2026-07-08T08:20:13+00:00urn:sha1:e9abe208bf5d1ffbbb5e8d6b0df187d97d89b1d9
Retro from the tasks-board build (see memory
claudomator-sequential-orchestration-gap): claudomator has no native way to
drive a stated intent through review-gated implementation to arbitrary
depth -- that role was played by a human/chatbot controller hand-dispatching
and re-dispatching all session. Design converged through several rejected
drafts (a separate "chain mode" pipeline, a distinct Planner role) before
landing on the actual shape: one role, one recursive decompose-or-evaluate
mechanism, applied at every depth, with arbitrated multi-evaluator review
running at every node instead of once at a story's outer level.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VTUSAEKfsPc6WGDq45yPHD