speccraft.validatelisted

Read `spec/commands/speccraft.validate.md` in full, then run final validation for the task provided. Input: read the arguments the user supplied after invoking this skill as an optional leading `auto` token, then the task id. `auto` present -> auto mode; absent -> interactive mode (default). Rules before starting (interactive mode): - confirm the task implementation is approved - confirm the task tests are approved - active phase must be execution - this command is read-only — no code changes allowed during validation - produce the Validation section (## Task: TASK-{id} → ### Validation) in spec/progress/<module>/progress-{STORY-ID}.md - check: implementation matches task scope, tests cover required behaviors, traceability chain is complete — 4-link (req → LLD → task → code → tests) for Layer: frontend tasks, 5-link (adds unit-test → integration-test) for Layer: backend tasks - after validation: output progress file path + pass/fail verdict + any gaps found, then STOP - mark task as done only if all checks pass and all gaps are recorded Rules before starting (auto mode — `auto` token present): - same confirmations and same read-only checks as interactive mode above — this command never writes code or tests, in either mode - there is no revise loop to self-review here (validation is read-only); apply `spec/commands/speccraft.validate.md` § Auto Mode instead: a `blocked` gap is logged to the traceability shard (`Approver: auto-mode-ai`, naming which earlier stage's own auto