uniface-procscript

name	uniface-procscript
description	Navigue et interroge la documentation de référence officielle Uniface 9.7 ProcScript (594 entrées couvrant les instructions, fonctions, triggers, types de données, directives préprocesseur et fonctions struct). À utiliser quand l'utilisateur pose des questions sur la syntaxe ProcScript, les triggers Uniface, les opérations base de données, la gestion des listes, la manipulation d'entités, les fonctions de chaînes, la gestion d'erreurs ou tout sujet de programmation Uniface 9.7.
version	3.1.1
license	Proprietary
metadata	{"author":"dedalus-erp-pas","uniface-version":"9.7","source-repo":"Dedalus-ERP-PAS/UNIFACE","source-branch":"DOC"}
allowed-tools	Read Grep Glob

Uniface 9.7 ProcScript Reference -- Navigation Skill

This skill gives you access to the complete official Uniface 9.7 ProcScript reference documentation -- 594 entries across 8 merged category files in reference/. Your role is to look up and read these files when answering ProcScript questions, not to rely on memorized knowledge.

What is ProcScript (minimal context)

ProcScript is the 4GL scripting language of the Uniface low-code platform. Key traits to keep in mind when reading the docs:

• Semicolons (;) start comments, not end statements
• One statement per line (%\ for continuation)
• Fields are referenced as FIELD.ENTITY (qualified) or FIELD (in context)
• Variables: $1-$10 (scratch), $NAME$ (general), variables/endvariables blocks
• Lists use gold-semicolon as separator
• $status holds the result of the last operation (0 = success, <0 = error)
• Code lives in triggers (event handlers) attached to components, entities, or fields

Documentation layout

All documentation is in reference/ relative to this skill. Each category is a single merged file containing all entries separated by --- horizontal rules:

reference/
  llms.txt                  # Index: every entry with title + one-line description (~620 lines)
  procstatements.md         # 173 entries -- statements (retrieve, read, store, if, for, putitem, ...)
  procfunctions.md          # 248 entries -- functions ($status, $concat, $scan, $replace, $date, ...)
  triggersstandard.md       # 74 entries  -- standard triggers (Read, Write, Store, Validate Field, ...)
  triggersextended.md       # 34 entries  -- extended triggers (OnClick, OnChange, expand, ...)
  procdatatypes.md          # 21 entries  -- data types (string, numeric, date, struct, handle, ...)
  procprecomp.md            # 23 entries  -- preprocessor directives (#define, #if, #include, ...)
  procstructfunctions.md    # 13 entries  -- struct functions ($name, $parent, $scalar, $tags, ...)
  predefinedoperations.md   # 8 entries   -- predefined operations (Exec, Init, Cleanup, Quit, ...)

Each entry starts with # Title and includes the full reference content: syntax, parameters, return values, usage notes, and code examples. Entries are separated by ---.

How to answer ProcScript questions

Follow this lookup strategy. Always consult the docs -- do not answer from memory alone.

Step 1: Identify what the user needs

Map the user's question to a category file:

User asks about...	Look in
A statement (retrieve, read, store, if, for, putitem, creocc, ...)	procstatements.md
A $function ($status, $concat, $scan, $date, $curocc, ...)	procfunctions.md
A trigger (Read, Write, Store, Validate, Leave Field, ...)	triggersstandard.md
An extended/widget trigger (OnClick, OnChange, expand, ...)	triggersextended.md
A data type (string, numeric, date, struct, handle, ...)	procdatatypes.md
Preprocessor (#define, #if, #include, compile-time constants)	procprecomp.md
Struct operations ($name, $parent, $scalar, $tags, ...)	procstructfunctions.md
Predefined operations (Exec, Init, Cleanup, Quit, ...)	predefinedoperations.md

Step 2: Find the entry within the file

Option A -- Grep for the entry heading (fastest, when you know the name):

Grep: pattern="^# retrieve$" path="reference/procstatements.md" -A 80

This returns the entry and the following lines. Read until you hit the next --- separator or # NextTitle.

Option B -- Search the index (when unsure of the exact name):

Read reference/llms.txt to find the entry name and its category file. Each section lists the file:

## Statements
File: `procstatements.md`

- retrieve: Activate the Read trigger for the first outermost entity...

Option C -- Grep across all files (for cross-cutting questions):

Grep: pattern="your search term" path="reference/" glob="*.md"

This is useful for questions like "which statements support the /lock option?" or "where is $webinfo used?".

Step 3: Read and synthesize

1. Use Grep with -A (after context) to read the entry. Typically -A 80 is enough for one entry; increase if the entry is longer
2. Pay attention to: syntax, parameters, return values, use restrictions, and code examples
3. Check the Related Topics section at the bottom of each entry for connected concepts
4. If the question spans multiple topics, grep multiple entries and synthesize
5. Always cite which file and entry you consulted

Step 4: Respond with precision

• Quote exact syntax from the docs
• Include code examples from the docs (they use ```procscript fenced blocks)
• Mention $status return values and error codes when relevant
• If the docs mention restrictions (component types, trigger contexts), include them
• When the user's question is ambiguous, list the possible interpretations with entry references

Common lookup patterns

"How do I do X?" (task-oriented)

1. Search reference/llms.txt for keywords related to the task
2. Grep the most relevant entries from the appropriate category file
3. Check reference/triggersstandard.md if the task involves a specific event context
4. Combine findings into a coherent answer with code examples from the docs

"What does X do?" (reference lookup)

1. Grep directly for the entry: pattern="^# X$" in the appropriate category file
2. Read and present: syntax, description, parameters, return values, examples

"What's wrong with this code?" (debugging)

1. Identify which statements/functions the code uses
2. Grep each one's entry, paying attention to:
   • Correct syntax (the docs show exact syntax templates)
   • Valid trigger contexts (some statements are only valid in specific triggers)
   • Return values and error conditions
   • Common gotchas mentioned in the "Use" or "Description" sections

"What are all the X?" (enumeration)

1. Read reference/llms.txt and filter by the relevant section header
2. Or Grep with a broad pattern across the relevant category file

Important caveats

• Generated files: The reference/ directory is regenerated by doc/scripts/convert.py from doc/html/ sources -- do not edit these files manually.
• Merged format: Each .md file contains multiple entries separated by ---. Use Grep with -A context to read individual entries.
• Encoding: Source files are Windows-1252; generated output is UTF-8.
• Version: This documentation covers Uniface 9.7 specifically. Do not assume features from other versions.