A small collection of skills for working in the Rexx ecosystem. Each skill is a self-contained Markdown document written for an AI agent (or a person) to follow: how to install something, how to run it, what traps to avoid.
Each skill names the exact versions it targets and the environment it assumes. Where a step is environment-specific, the skill says so — adjust those to your own machine.
oorexx — Pitfalls and best
practices for ooRexx code. Covers the special variables that get
silently overwritten (RESULT, RC,
SIGL), non-short-circuit logical operators, method scoping
with Expose, where labels are illegal, and idioms for
strings, stems, and line endings.
Works with: Claude.
oorexx-run — Install and
run ooRexx. Pinned to the 5.3.0beta branch; covers downloading the right
.deb from SourceForge, verifying it, the
::Requires lookup order, and the common errors.
Works with: Claude.
apache-cgi-oorexx
— Serve ooRexx scripts over CGI with Apache. Covers installation, the
two configuration patterns (running .rex scripts directly
and processing .md files through a Rexx handler), a shell
wrapper for the CRLF-shebang trap when scripts are edited on Windows,
and Linux case-sensitivity gotchas.
Works with: Claude.
rexxhttp — Serve ooRexx
servlets over Apache with the
RexxHTTP processor. Requires apache-cgi-oorexx (and
through it oorexx-run); covers
a basic sample installation, the Apache wiring, end-to-end verification
with curl, and the traps that bite when writing servlets (the frozen
invocation contract, non-short-circuit &, the JSON ::Requires,
case-sensitivity).
Works with: Claude.
ebnf2svg — Generate one SVG railroad diagram per production from a W3C-style EBNF grammar. Covers the toolchain (the Java-based Railroad Diagram Generator rr.war plus two ooRexx scripts), where to download each piece, prerequisites, usage and the no-extension argument convention, the EBNF notation, and how the orchestrator chains generation and SVG extraction.
Works with: Claude.
oorexx-docbook-build
— Build the official ooRexx documentation books (DocBook + Publican
layout + Apache FOP) to PDF, with optional Rexx syntax highlighting on
the program listings. Requires oorexx-run, and ebnf2svg only if a book builds
railroad diagrams from EBNF. Covers checking out the sources over
svn-https, the official build scripts plus the Rexx Parser's drop-in
highlighting tools (hldocprep/hldoc2pdf), the
docpath layout for a book of your own, and the traps that
bite (the wrong highlight.rex shortcut, re-highlighting a
clean checkout).
Works with: Claude.
regina-run — Install and run Regina, the ANSI-standard classic Rexx interpreter. Covers installing from the distro package manager, invoking it, and getting a newer build from SourceForge if you need one.
Works with: Claude.
apache-cgi-regina — Serve Regina Rexx scripts over CGI with Apache. Covers installation, the two configuration patterns (running .rex scripts directly and processing .md files through a Rexx handler), loading external routines with CALL and the REGINA_MACROS search path, a shell wrapper for the CRLF-shebang trap when scripts are edited on Windows, and Linux case-sensitivity gotchas.
Works with: Claude.
Open the one you need and follow it top to bottom. They assume basic command-line comfort and, for most, a working Rexx interpreter — so if you don't have one yet, start with oorexx-run or regina-run.