implementation-harness-add-to-backloglisted

# Author / extend the .harness/tracking/TASKS.json backlog You are turning a feature or phase description into well-formed `.harness/tracking/TASKS.json` task objects that the single-loop harness can build. Read this whole file, then execute in order. The cardinal rule: **append, never clobber** — existing *pending / in-flight* task objects and their `status` are sacred. (Completed `done` tasks may be deliberately pruned to keep the backlog readable — see §7 — but are never silently altered during an append.) ## 1. Pre-flight - Require the harness: `.harness/tracking/TASKS.json`, `.harness/docs/HARNESS.md`, and `.harness/scripts/loop.sh` must exist in the project. If any is missing, stop and point the user at `implementation-harness:implementation-harness-create` first. (Either loop variant — worktree or in-place — installs as `.harness/scripts/loop.sh` and keeps `.harness/tracking/TASKS.json` at the same fixed path, so this skill is identical for both.) - Require `jq` (the loop and this skill use it). If absent, tell the user to `brew install jq`. - **Read `.harness/docs/HARNESS.md` §8.1** (the task schema) live — bind to the actual schema in this project, in case it has evolved. Don't rely on a hardcoded copy. - **Read `.harness/tracking/TASKS.json`** and extract, with jq: - the highest existing id — `jq -r '.tasks[].id' .harness/tracking/TASKS.json | sort | tail -1` → new ids continue monotonically, zero-padded to the same width (≥3 digits); - all existi