follow-up

Run the follow-up status check, which sweeps the follow-ups directory for ACTIVE entries and reports each one's disposal classification. If none are active, it falls through to open risk-register and tech-debt rows plus value-rubric-ranked plans and findings.

The check synthesizes one recommendation in priority order: any ACTIVE follow-up first, then the value-rubric's top tier-1 item, then a labeled qualitative judgment call across open register and tech-debt rows. It always states why the pick beats the runner-up.

No, the check is strictly read-only and never mutates the register, tech-debt list, or any follow-up's status. It proposes the action and hands off to the risk-register, tech-debt, or plan-create skill only after you confirm.

An empty follow-up list is the expected steady state once items are promoted or closed, so the real backlog usually lives in open register and tech-debt rows. The check always falls through to those sources rather than reporting an empty list as the whole answer.

No, register and tech-debt rows carry no value field and were never graded against the value-rubric protocol. A pick sourced from them is reported explicitly as a qualitative judgment call, not rubric-sorted output.

The check compares the index's header entry count against the actual file count under the context directory before trusting it. On a mismatch, it regenerates the index via the context-index skill before building the candidate pool.