executing-planslisted

# Executing Plans Autonomous task execution with QA gates across multiple AI platforms. ## Overview The AI Development Loop: - AI agents pick up and implement tasks from a GitHub Issues queue - You do QA only (approve or reject issues in the Human Review column) - Multiple platforms (Claude CLI, Cursor, Codex) can work in parallel - Switch between platforms to maximize rate limits ## Architecture ``` ┌─────────────┐ ┌─────────────┐ ┌──────────────┐ ┌─────────────┐ │ BACKLOG │───▶│ IN PROGRESS │───▶│ HUMAN REVIEW │───▶│ DONE │ │ │ │ │ │ │ │ │ │ open + gate │ │ Agent picks │ │ YOU review │ │ Shipped │ │ (opted in) │ │ & builds │ │ the PR (you) │ │ (closed) │ └─────────────┘ └─────────────┘ └──────────────┘ └─────────────┘ │ │ ┌─────��─────┐ ┌──────┴──────┐ │ Claude │ │ Reject │ │ Codex │ │ → Backlog │ └───────────┘ └─────────────┘ loop:planning→executing→testing→shipping (labels) (Deferred = parked / wontfix, out of the main flow) ``` Columns map to GitHub Issues state + the board `Status` field — the **sole source of truth** for where an issue sits. There are no `status:*` labels: | Column | Issue state | Board `Status` | | ------------ | ----------- | -------------- | | Backlog