Skill

debug-deployment

Debugs failed or misbehaving dltHub Runtime deployments by checking job status and logs, canceling runs, accessing production data, opening dashboard, and diagnosing common issues like missing dependencies or secrets.

Python

Bash

deployment

devops

npx claudepluginhub dlt-hub/dlthub-ai-workbench --plugin dlthub-runtime

Tool Access

This skill uses the workspace's default tool permissions.

Preview

**Reference**: https://dlthub.com/docs/hub/runtime/overview.md

SKILL.md

Similar Skills

deploy-workspace

Deploys dlt pipelines to dltHub Runtime. Prepares Python scripts by removing dev_mode/limits, verifies dispositions, pins versions; runs dlt CLI deploy/launch/logs. For runtime deployment requests.

dlthub-runtime

render-debug

Debugs failed Render deployments by analyzing logs, metrics, and database state. Identifies errors like missing env vars, port binding, OOM and suggests fixes for service crashes, health check failures, or performance issues.

6 files

render

debug-pipeline

Debugs and inspects dlt pipelines after runs, checking traces, load packages, schemas, data, and errors like missing credentials or failed jobs. Use post-execution.

rest-api-pipeline

Stats

Parent Repo Stars19

Parent Repo Forks2

Last CommitMar 2, 2026

Actions

View Source View Plugin View on GitHub View README

Help us improve

Share bugs, ideas, or general feedback.

Debug dltHub Runtime deployment

Reference: https://dlthub.com/docs/hub/runtime/overview.md

Check job status

dlt runtime job list                              # all jobs
dlt runtime job info <script_or_name>             # details for one job
dlt runtime job-run list <script_or_name>         # list all runs
dlt runtime job-run info <script_or_name> [run#]  # specific run details

View logs

dlt runtime logs <script_or_name>                 # latest run
dlt runtime logs <script_or_name> <run#>          # specific run

Note: logs always follow. You will get stuck if job is still running. This is a temporary limitation.

Cancel a running job

dlt runtime cancel <script_or_name> [run#]

Access production data (read only)

Figure out right profile for data access.

access profile if it is configured (list profiles). if not: prod profile (if configured)
if none is present ask user which profile to use

ALWAYS ask human before accessing production data. Confirm the profile
pin the profile
use mcp tools, run cli, python scripts
pin dev profile after work is done

to run a single command on given profile use:

WORKSPACE__PROFILE=prod dlt pipeline my_pipeline info

Note: you must pin profile for mcp server to see the change

Open the web dashboard (for humans)

dlt runtime dashboard

Opens the dltHub Runtime UI at dlthub.app — shows jobs, runs, logs, schedules, and deployment history.

Quick diagnosis

If a job failed:

dlt runtime job-run info <script> <run#> — check exit status and timing
dlt runtime logs <script> <run#> — read the error output
Common causes:
- Missing dependencies in pyproject.toml — all packages must be declared, not just locally installed
- Secrets not configured for prod profile — runtime uses prod profile, check .dlt/prod.secrets.toml
- Script missing if __name__ == "__main__": — the job does nothing without it
- dev_mode=True left in — drops and recreates dataset on every run
- Wrong destination credentials — prod profile may point to a different destination than dev
- Job timeout — jobs are killed after 120 minutes; use incremental loads for long pipelines
After fixing, redeploy with dlt runtime deploy and relaunch with dlt runtime launch <script>