Uh oh!
There was an error while loading. Please reload this page.
docs(prompts): tell Prime Claude that .dict and .kbb are the KB source files - #207
Merged
ddehilster merged 1 commit intoAug 21, 2026
Merged
Conversation
…e files Priming a fresh session on this prompt sent Claude straight into the Knowledge Base template's kb/user/*.kb files (attr.kb, hier.kb, phr.kb, word.kb) to "study the knowledge base" — those are the engine's runtime dump, not files anyone authors. hier.kb alone is 33 KB, so it is the first thing an exploring agent finds. Say it plainly instead: kb/user holds the KB *sources*, and .dict and .kbb are the only KB files to read or write. Name the .kb files as engine output that must not be read as examples or hand-edited, and repeat the point in the engine-facts bullet, which now also describes what a .kbb is and how it pairs with the .dict of the same stem. Also folds in the pending "build on the libraries" rules of thumb: look in the language/misc libraries before hand-typing a vocabulary, wire one in by copying the .dict/.kbb pair into kb/user, match on dictionary attributes rather than words, match _phrase as well as _xALPHA, join on a stable attribute rather than $text, and extend the library file (or its generator) instead of hard-coding missing words in a pass. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Uh oh!
There was an error while loading. Please reload this page.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for freeto join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Priming a fresh Claude session on Prime Claude for NLP++ sent it straight into the Knowledge Base template's
kb/user/*.kbfiles —attr.kb,hier.kb,phr.kb,word.kb— to "study the knowledge base." Those are the engine's runtime dump, not files anyone authors, andhier.kbalone is 33 KB, so it is the biggest and first thing an exploring agent finds in that folder.Changes
kb/user/holds the knowledge base sources, states that.dictdictionaries and.kbbknowledge base files are the only KB files to read or write, and names the four.kbfiles as engine output — not to be read as examples, copied, or hand-edited..kbbis (indented concept hierarchy with attributes, paired with the.dictof the same stem) and repeats that.kbfiles are engine output.Also folds in the pending "build on the libraries, never on hand-typed word lists" rules of thumb that were sitting uncommitted on the merged
feat/english-nationalitiesbranch: look in the language/misc libraries before hand-typing a vocabulary, wire one in by copying the.dict/.kbbpair intokb/user/, match on dictionary attributes rather than words, match_phraseas well as_xALPHA, join on a stable attribute rather than$text, and extend the library file (or its generator) instead of hard-coding missing words in a pass.Prompt text only — no engine, analyzer, or library files touched.
🤖 Generated with Claude Code