karvey-taskslisted

# Karvey Tasks ## Purpose Generate the implementation task plan from the approved architecture. Record it in ClickUp (Epic > Feature > Tasks with dependencies) or in a PLAN.md checklist. Target size: 10–30 minutes per task (AI timings). ## Execution steps ### Step 1 — Load context Read: - `docs/spec/changes/{change-id}/spec.json` - `docs/spec/changes/{change-id}/requirements.md` - `docs/spec/changes/{change-id}/architecture.md` - `docs/spec/changes/{change-id}/infra.md` - `rules/clickup-protocol.md` Verify `approvals.infra.approved = true`. If not, stop. Determine sequential mode: if `--sequential`, do not use parallelism markers. ### Step 2 — Generate tasks.md draft For each Feature identified in architecture.md, generate tasks per layer following the File Structure Plan. **Granularity rules:** - 1 task = 1 unit of work executable by an AI agent - Estimated time: 10–30 minutes (AI timings) - Maximum 1h per task. If it exceeds that, split it. - Each task produces a verifiable artifact (SP, endpoint, component, test) - The "done" criterion must be observable (not "implement X", but "the SP works and returns {Y}") **Mandatory execution order:** ``` [DB] → [Backend] → [Frontend] ``` Tasks of the same layer within a Feature can be marked `(P)` if they are independent. **Valid layer labels:** `[DB/Backend/Frontend/Infra]`. `Infra`-type tasks are allowed for IaC/pipeline adjustments that come up during implementation (the base infra is already defined in `infra.md`). *