Skills
A skill is a procedure your agent runs: how to prepare a data pile, author a world, review an existing state, or classify a corpus at scale — with the governance steps built in. They ship with the repo; hand one to your agent and point it at your data.
classification-at-scale
SKILL.md →Classify an internal catalog against a standard taxonomy at scale using deterministic providers, tri-state signals, batch group review, and a trust flywheel — LLM judgment only for configs, ambiguous tails, and critique.
create-state
SKILL.md →Create a new Cruxible state from raw domain data through staged graph, workflow, query, and review-loop design.
overlay-and-fit
SKILL.md →Create an overlay from a published reference state, use the inherited state plus any applied kit with the lowest-friction path, and only add local config or code when the current overlay is not enough.
prepare-data
SKILL.md →Profile and prepare raw source files before modeling or loading them into Cruxible; validate keys, grain, joins, and transformation needs, then produce a concrete readiness report.
review-state
SKILL.md →Review an existing state, inspect evaluation, queries, governed groups, feedback, and outcomes, and surface prioritized issues with the likely fix surface before making changes.
- ·finding structural or behavioral problems
- ·inspecting representative queries and receipts
- ·reviewing governed groups and their outputs
- ·spotting repeated feedback or outcome patterns
- ·telling the user what is wrong and what should be fixed next
wiki-to-state
SKILL.md →Convert an informal LLM wiki (CLAUDE.md files, a memory-bank directory, an Obsidian vault) into governed Cruxible state through register/propose/review stages, with every claim citing the pinned wiki chunks as evidence.