help(prompts): repurpose build-analyzer as "Prime Claude for NLP++"

Help/markdown/vscode/prompts/00-general-pointers.md

@@ -1,7 +1,9 @@
 General pointers: Claude + NLP++
-<!-- desc: Start here — how Claude helps with NLP++ work (gathering development texts, mimicking the rule/code patterns of a logical language, and assisting on an existing analyzer), plus a guide to the four prompt files in this library and when to reach for each. -->
+<!-- desc: Start here — how Claude helps with NLP++ work (gathering development texts, mimicking the rule/code patterns of a logical language, and assisting on an existing analyzer), plus a guide to the five prompt files in this library and when to reach for each. -->
 Claude pairs well with NLP++. This page is an overview; the other entries in this list are ready-to-use prompts you open, fill in, and paste into Claude.
 
+**Start every new Claude session by pasting in [Prime Claude for NLP++](02-prime-claude.md).** It is everything Claude needs to know to work on your analyzers — the installed engine, example, and template paths, the KB and JSON conventions, and the engine facts that trip up newcomers. Paste it first, let Claude read the examples and templates, and only then ask for whatever you need. The other prompts below are task-specific and assume Claude has already been primed this way.
+
 Why the pairing works:
 
 - **Claude assists across the whole workflow of an existing analyzer.** Beyond building from scratch, it can add or refine grammar passes, write dictionaries and knowledge bases, generate more test inputs, debug why a rule does or doesn't fire, and explain unfamiliar passes or library functions.
@@ -12,7 +14,7 @@ The five prompt files in this library:
 
 1. **[From scratch: chemical formulas](01-chemical-formulas.md)** — A complete worked example. It has Claude gather a Wikipedia chemistry corpus and build an analyzer that finds chemical formulas and breaks each into its elements and atom counts, emitting JSON. Use it to watch an analyzer built end to end, or as a model for your own from-scratch build.
 
-2. **[NLP++ local install — build an analyzer](02-build-analyzer.md)** — The generic starting point for a brand-new analyzer. It hands Claude the installed engine, example, and template paths plus the conventions that keep it on the rails, then leaves a blank for you to describe the corpus and the extraction task.
+2. **[Prime Claude for NLP++](02-prime-claude.md)** — Paste this into a fresh Claude session before anything else. It is everything Claude needs to know to work on your analyzers: the installed engine, example, and template paths, the conventions that keep it on the rails, and the engine facts newcomers stumble on. There is no blank to fill in — once Claude has read it and studied the examples, describe whatever task you have.
 
 3. **[Harden analyzer](03-more-text-files.md)** — For an analyzer you already have. It generates additional, varied test inputs (including edge cases), runs them through the engine, and reports where the extraction looks wrong so you can tighten rules and re-bless your regression goldens.
 

Help/markdown/vscode/prompts/02-build-analyzer.md → Help/markdown/vscode/prompts/02-prime-claude.md

@@ -1,6 +1,6 @@
-NLP++ local install — build an analyzer
-<!-- desc: Generic starting prompt — hands Claude the installed engine, example, and template paths, plus the conventions that keep it on the rails (use the Knowledge Base template as intended, build results into a KB and emit JSON with JsonKB, and run with -WORK pointing at the engine). Fill in your task at the bottom. -->
-I want you to help me write a prototype NLP++ analyzer. NLP++ is a rule-based programming language for natural language processing, run by the NLP engine. Everything you need is already installed on this machine at the paths below:
+Prime Claude for NLP++
+<!-- desc: Paste this into a fresh Claude session before asking for any NLP++ help — it is everything Claude needs to know to work on your analyzers: the installed engine, example, and template paths, the conventions that keep it on the rails (use the Knowledge Base template as intended, build results into a KB and emit JSON with JsonKB, and run with -WORK pointing at the engine), and the engine facts that trip up newcomers. No task to fill in — once Claude has read this, ask it for whatever you need. -->
+This is everything you need to know to help me work on NLP++ analyzers. NLP++ is a rule-based programming language for natural language processing, run by the NLP engine. Read this in full before we start — afterward I will ask you to build a new analyzer, extend an existing one, write dictionaries and knowledge bases, generate test inputs, or debug why a rule fires. Everything you need is already installed on this machine at the paths below:
 
 - NLP engine executable (run analyzers with this): {{engineExe}}
 - NLP engine command-line switches (how to run, compile, set the working directory, etc.): https://github.com/VisualText/nlp-engine/blob/master/README.md#switches
@@ -29,6 +29,4 @@ Run an analyzer with:
 
 -WORK must point at the engine directory (the folder containing nlp.exe), NOT the analyzer folder — otherwise the engine prints "ERROR IN ANALYZER INIT", silently runs only the tokenizer, and no passes execute. Per-input output (and .tree debug files under -DEV) appears in input/<file>.txt_log/.
 
-Study a couple of the example and template analyzers first to learn the conventions. Then build the passes, run the analyzer over the input files, and show me the output so we can check it together.
-
-Create a folder of text files from the internet that (FILL IN YOUR DESCRIPTION) and create an analyzer that does (FILL IN YOUR DESCRIPTION OF THE ANALYZER).
+Study a couple of the example and template analyzers now to learn the conventions, then tell me you are ready and I will describe the task. When we work, build the passes, run the analyzer over the input files, and show me the output so we can check it together.