ai-engineering-updatelisted

# AI Tooling Refresh The currency engine behind `ai-engineering`. It discovers new AI tooling, verifies what changed, and folds findings into the catalog **deterministically and without duplicates**. The hard parts - dedupe, classification, conflict resolution, freshness - are handled by `scripts/ledger.py` in the `ai-engineering` skill; this skill is the research-and-curate workflow around it. ## Why this exists The corpus is maintained so that answers about AI tooling stay ahead of the model's training data. Each refresh pass is what buys that, which is why every entry carries a date and, where the project publishes one, a version. The source catalog also grew by people repeatedly "finding" the same repo and pasting it again. The ledger ends that: every URL has a canonical form and a record. Before any link is added, it is checked against the catalog. New knowledge accretes in **data** (`catalog.tsv`, `seed-sections.tsv`, `rules.tsv`), never by editing code, so it scales without limit. **Read `resources/data-contract.md` in the `ai-engineering` skill before adding a file, a column or a store.** It states the canonical-URL join, which files are generated, the tag vocabulary and its legacy spelling, and the conventions a new store has to honor. Writing a store that does not join to the catalog produces an island nothing else can see. ## The engine (read before using) The `ai-engineering` skill's `scripts/ledger.py` (next to its SKILL.md; glob `**/skills/ai-engineering/