ai-asset-pricing: Skill for Claude Code

WRDS SSH Connection Skill

This skill provides guidance on connecting to and working with WRDS (Wharton Research Data Services) servers via SSH.

Examples

• /wrds-ssh -- connect to WRDS and explore the server
• /wrds-ssh submit analysis.sas -- upload and submit a SAS job
• /wrds-ssh download ~/scratch/output.csv -- transfer results back

Connection

The SSH connection to WRDS is pre-configured. Connect using:

ssh wrds

This connects to the WRDS cloud server (wrds-cloud-sshkey.wharton.upenn.edu) via the ~/.ssh/config host alias.

WRDS Paths

• Home directory: ~ (on WRDS)
• Scratch directory: ~/scratch (symlink — see CLAUDE.md setup instructions)

All SSH commands use single quotes to avoid local shell expansion. Paths like ~/scratch/ expand on the remote side.

Running Commands on WRDS

Interactive Session

For quick commands or testing:

ssh wrds "command_here"

Running SAS Code

IMPORTANT: On WRDS Cloud, you cannot run sas directly. You must use qsas which submits jobs to the Grid Engine queue. Jobs run asynchronously.

Step 1: Create the SAS program file on WRDS:

ssh wrds 'cat > ~/scratch/myprogram.sas << EOF
options nonotes nosource;
proc contents data=crsp.dsf; run;
EOF'

Step 2: Submit the job:

ssh wrds 'qsas ~/scratch/myprogram.sas'

This returns immediately with a job ID (e.g., Your job 12345678 ("myprogram.sas") has been submitted).

Step 3: Wait for completion and check output:

# Check if job is still running
ssh wrds 'qstat -u $(whoami)'

# Once complete, check the log for errors
ssh wrds 'tail -50 ~/scratch/myprogram.log'

# View the listing output
ssh wrds 'cat ~/scratch/myprogram.lst'

Output files: SAS creates .log and .lst files in the same directory as the .sas file.

NOTE: The -sync y option does NOT work on WRDS Cloud. Jobs are always asynchronous. Use sleep or poll qstat to wait for completion.

Running SQL Queries (PostgreSQL)