plain-words

Sketch data shapes (grain, keys, schemas), function signatures, and the module or pipeline structure before code, then stay in the loop while implementation…

Help a user pick the right astack skill, analyst-mode playbook, or principle for a data analysis task, such as EDA, a metric move, a notebook, a pipeline, or…

Vet a runtime measurement for a query, dataframe transform, pipeline, or notebook before you report or act on it: limiter, tuning, data volume, warm vs cold…

Find what a change to a table, column, transform, metric, or notebook could break downstream (dashboards, saved queries, notebooks, consumers) before it…

Data analysis style for an analysis repo.

Spawn the comment-critic agent on a data repo's code, notebooks, or SQL, act on its accepted findings, and offer encodings for claimed constraints.

Catch the user up on what changed in this chat since their last message: pipeline runs, numbers, PRs, and agents, in a few plain sentences and small visuals.

Run N parallel candidates at the same analysis, pipeline, notebook, or chart task, pick a base, and graft the strongest parts of the rest into it.

Apply at the start of an analysis, before writing the first query.

Apply when you write the same instruction a second time, or notice a recurring correction, such as a join that drops rows, a timezone mix-up, or a chart axis…

Apply when wiring data ingestion, parsing raw files, validating input schemas, or connecting to databases, APIs, and notebooks.

Apply to any non-trivial analysis or data work: bulk edits, notebook conversions, backfills, reconciliations, and checks.