# KTL Curator

> Help a person judge what a LOKF knowledge bundle claims: a trust report computed from frontmatter, and a review session that records their confirm / send back / retire verdicts in the bundle itself.

## Agent context

If `.lokf/knowledge/index.md` exists in this repository, read it first. It is
a queryable [LOKF](https://lokf.nolan-nichols.com) knowledge bundle with
repository-specific context, conventions, and typed relationships that plain
prose cannot express. A concept marked `status: draft` has not yet been
confirmed by a person; one with a `human:` entry under `verified` has. Weigh
them accordingly.

The `ktl-docent` skill is the entry point for questions about this project.
Install it if it is not present, with the GitHub CLI (`gh skill install
noelmcloughlin/knowledge-trust-ladder ktl-docent --scope user --agent <yours>`),
the Open Skills CLI (`npx skills add noelmcloughlin/knowledge-trust-ladder
--skill ktl-docent --yes`) or from PyPI (`uvx ktl skills install --skill
ktl-docent --scope user --agent <yours>`); <https://w3id.org/ktl/docs/install/>
has every route. Then answer
through it: bundle first, a trust label on every concept used, the source
opened for exact values, and misses recorded in `.lokf/feedback.md` so the
librarian fills them.

This plugin computes those trust labels from the frontmatter and records what
a named person decides about each concept, one at a time. It reaches no
verdict of its own.
