Skip to content

SKILL.md

Latest commit

 

History

History
81 lines (63 loc) · 3 KB

SKILL.md

File metadata and controls

81 lines (63 loc) · 3 KB
Raw
name submitting-jobs
description Submits Wherobots Cloud job runs from local Python files or S3 URIs. Resolves storage using the customer's S3 Storage Integration first, falling back to Wherobots managed storage. Use when a user wants to run, submit, or execute a Python script on Wherobots Cloud.

Submitting Wherobots Job Runs

Storage resolution order

Job runs execute from S3. The CLI accepts a local .py file or an s3:// URI — local files are uploaded automatically. Resolve where the script will live in this order:

1. Customer's own S3 via Storage Integration (preferred)

If the customer has an S3 Storage Integration configured:

  • Direct s3:// URI: s3://their-bucket/scripts/my_job.py
  • Upload override: --upload-path s3://their-bucket/prefix
  • Environment variable: WHEROBOTS_UPLOAD_PATH

2. Wherobots managed storage (fallback)

If no Storage Integration exists or the customer is unsure, the CLI automatically resolves the managed storage bucket via the organization API. No extra configuration needed — just pass the local .py file.

Command reference

wherobots job-runs create <script> -n <name> [flags]

Required

  • <script> — local .py file path or s3:// URI (positional)
  • -n, --name — job run name

Optional

  • -r, --runtime — compute size (default: tiny)
  • --run-region — region (default: aws-us-west-2)
  • --timeout — seconds (default: 3600)
  • --args — space-separated arguments for the script
  • -c, --spark-config — key=value (repeatable)
  • --dep-pypi — name==version (repeatable)
  • --dep-file — S3 URI for .jar, .whl, .zip, .json (repeatable)
  • -w, --watch — stream logs until completion
  • --upload-path — override S3 upload destination
  • --no-upload — skip auto-upload (script already in S3)
  • --output — text (default) or json

Examples

Local file, managed storage:

wherobots job-runs create ./my_script.py -n my-job -w

Local file, customer S3:

wherobots job-runs create ./my_script.py -n my-job --upload-path s3://my-bucket/jobs -w

Script already in S3:

wherobots job-runs create s3://my-bucket/scripts/my_script.py -n my-job --no-upload -w

With dependencies and Spark config:

wherobots job-runs create ./etl.py -n nightly-etl \
  -r medium \
  --timeout 7200 \
  --dep-pypi pandas==2.1.0 \
  --dep-file s3://my-bucket/libs/utils.whl \
  -c spark.sql.shuffle.partitions=200 \
  -w

Guidance

  • This skill is for batch job execution. If the user wants to run an interactive query, use the MCP server or Python DB-API instead — see the using-wherobots skill.
  • Ask whether the customer has an S3 Storage Integration or should use managed storage.
  • Default to -w (watch) so they see logs in real time.
  • Only ask about runtime size if they mention performance needs or large data.
  • Only ask about dependencies if they mention libraries their script needs.
  • WHEROBOTS_API_KEY must be set. The wherobots CLI must be installed.