CawPlan Knowledge

Bootstrap

cawplan skill check

Navigation Model

The whole knowledge base is one continuous drillable tree: datasets → documents → headings → section content, plus search results, which are the same tree pre-filtered by relevance. The user never needs to know or type CLI flag names (--outline, --section, --dataset, --grep, ...) — those are implementation details you choose based on what level of the tree they're asking about. If a user pastes a literal cawplan ... command instead of a plain-language request, treat it as identifying what they want (a dataset, a document, a section), not as a script to execute verbatim; still pick the right underlying flags yourself.

At every level, decide the presentation by candidate-set size and how iterative the pick will be: