Spec Readiness Check

The spec — copy this

Set up a new bot for me I can trigger against any spec from this directory. Walk me through connecting my case management system, then configure it: read the spec I point it at, extract every field and data condition it depends on — lead source, incident date, jurisdiction, matter type, cost categories, activity timestamps, disposition reasons, whatever the spec names — and check each one against my live system: does a field exist for it, what is its fill rate, how consistent are its values, and how far back does usable history run. Return a verdict in three buckets: ready to run; will run but the output will be unreliable, and here is exactly why; and cannot run until this is fixed. Give the specific remediation for every gap and a rough sense of the effort. Where a spec needs history to mean anything — anything comparing cohorts, trailing medians, or year-over-year — tell me whether I have accumulated enough of it yet, because a correct spec run against four weeks of data produces a confident wrong answer. Never report a spec as ready on the basis that a field exists; existence is not fill rate. Ask me which spec, which system, and my minimum acceptable fill rate, run it against a spec I have already tried so I can compare its verdict to what actually happened, then save it.