Ordewell uses a read-only coding agent as planner, then gates each task on a marker-based verdict.

Ordewell turns one goal into an ordered plan where each task gets its own runner, model, thinking-effort level and execution mode, executed across a dependency graph. The planner is itself a coding agent (Claude Code, Codex or OpenCode) running read-only over the workspace, interleaving clarifying questions with research in one persistent conversation, which sidesteps needing separate API credentials. The VerdictEngine only completes a task once its marker appears, so a session that exits without one fails visibly instead of passing silently. 104 stars, 91 commits.