Existing Skills Made Better

helpme

In the helpme skill, standard error text produced on a command failure is searched against data from the Unix & Linux Stack Exchange forum to find an accepted answer that solves the issue, or the most highly rated answer in the case of an accepted answer being unavailable. This employs a simple index-based text search to identify the relevant Stack Exchange post and the corresponding answer.  The body of the result is then compared against available Linux manpages using a REST API from CLAI's man page explorer skill to recommend a relevant command to explore.

This feature has the potential to be incredibly helpful to USS users, as the proprietary and often esoteric nature of IBM Z can make it difficult to simply search the Internet for a solution to any problems that may arise.  But this fact also limits the usefulness of helpme in its original form: by searching the Unix & Linux Stack Exchange forums, helpme is likely to return information that doesn't apply to USS or z/OS.  To address this, our team expanded helpme's search functionality to default to the z/OS KnowledgeCenter when run from USS.  Specifically, standard error is sent as a query to the z/OS KnowledgeCenter API with the tags for the USS product applied.  USS publications are searched first, then Redbooks, then DeveloperWorks articles, and then finally Technotes.  Unlike the Stack Exchange search, this is very simplistic: the first result found in any KnowledgeCenter database query is the result used for the manpages comparison and eventually returned to the user.

nlc2cmd

nlc2cmd allows users to specify tasks in English and retrieve Linux command line syntax with the aid of Watson Assistant-based natural language classifiers.  To support USS, the core CLAI team at IBM Research created a new classifier with the ability to identify eleven different USS commands from a natural language description.  As a result, nlc2cmd can not only respond to USS-specific phrases like "copy file to pds member", but can also respond to phrases such as "edit a file" with suggested USS commands instead of their Linux counterparts.

Bringing the Power of AI to the Z Command Line

While the new skills are the flashiest part of the work that the zATT did with CLAI, they are far from all that we accomplished.  Our team worked on a number of low-level fixes to the base CLAI project, modified CLAI to be able install and run on USS, improved the overall CLAI installation process, added a new user installation option for non-root users, and disabled on USS any CLAI skills whose dependencies we could not port.  We feel that the end result is a powerful demonstration of the kinds of improvements that artificial intelligence can bring to USS, and hope that it will serve as a springboard for future work in applying AI to command-line interfaces on IBM Z and its operating system families.

Trying it out for yourself

The best part of using CLAI on z/OS is that anybody with their own Unix Systems Services ID can try it out for themselves! Here's how to do it:

1. Make sure that your z/OS system has the following prerequisites; if not, you may need to contact your system administrator.  Software requirements are available either through IzODA or are downloadable from Rocket Software.

   1. Bash 4.3 or higher 

   2. Python 3.6 or higher 

      • For best results, use IBM Open Enterprise Python for z/OS 3.8

      • Rocket Python 3.6 (both the standard release as well as the version available through the IzODA Anaconda stream) work as well, but with some caveats (documented here)

   3. Git 

   4. The path to the env program is /usr/bin/env

      • If this is not the case, your system administrator will need to create a symbolic link: ln -s /bin/env /usr/bin/env

2. If your environment is not already configured for Bash and Python:

   1. Configure your .profile

      export SHELL=path_to_bash
      exec $SHELL --login

   2. Configure your .bash_profile to launch $HOME/.bashrc

   3. Configure your .bashrc to add Python to system paths
      • Add Python to system paths

        export PYTHON_HOME=/path/to/python
        export LIBPATH=$PYTHON_HOME/lib:$LIBPATH
        export FFI_LIB=$PYTHON_HOME/lib/ffi
        export MANPATH=$PYTHON_HOME/share/man:$MANPATH
        export PKG_CONFIG_PATH=$PYTHON_HOME/lib/pkgconfig:$PYTHON_HOME/share/pkgconfig
        export CURL_CA_BUNDLE=$PYTHON_HOME/etc/ssl/cacert.pem

      • Respect file tags for ASCII/EBCDIC conversion
        export _BPXK_AUTOCVT=ON
        export _CEE_RUNOPTS="FILETAG(AUTOCVT,AUTOTAG) POSIX(ON)"

      • Set TERMINFO so that vi works properly:
        export TERMINFO=/usr/share/lib/terminfo

      • If your z/OS system employs multiple TCP/IP stacks, specify the name of the particular TCP/IP stack to be used
        export _BPXK_SETIBMOPT_TRANSPORT=TCP342

   4. Restart your shell (log out and then back in again)

3. Clone the CLAI git repository:
   git clone git@github.com:IBM/clai.git

4. Follow the notes in known-issues.md with regards to your Python configuration

5. Install CLAI

   cd path_to_zclai_repository
   make clean
   • To have the CLAI server use the default port number of 8010:
     make install
   • If you wish to have CLAI use a specific port number:
     ./install.sh --user --port=8nnn

6. Restart your shell

7. Use CLAI!

   1. Issue clai skills to see what skills are available for USS

   2. Use clai activate skillname to activate a skill

   3. Use clai deactivate skillname to deactivate a skill

   4. Issue clai stop to stop the CLAI server, and clai start to start the CLAI server

8. If you want to uninstall CLAI:

   cd path_to_zclai_repository
   make uninstall

Further Reading

• The CLAI Project Homepage 
• CLAI Development on Github 
• IBM Research Blog on CLAI 
• CLAI skills:
  • fixit 
  • helpme 
  • howdoi 
  • man page explorer 
  • nlc2cmd 
• CLAI skills for z/OS:
  • linuss
  • zmsgcode