A useful chart starts with a question and a trustworthy dataset. Ask your agent to explain the data’s shape and limitations before choosing a cleaning rule or calculating a trend.
Prepare the task
Keep a copy of the original data. Provide column definitions, units, time zones, and the question you want answered in the agent’s workspace. Use a non-sensitive task description when searching for skills.
Work through it with your agent
Inspect before changing
Find a skill for dataset profiling or analysis. Ask the agent to report missing values, duplicate keys, inconsistent types, and plausible ranges without silently deleting records.
Make transformations explicit
Approve a cleaning policy and record the rows affected by each operation. Preserve the source data and generate a repeatable script or notebook for the transformation.
Explain the result
Select summaries and charts that answer the original question. Show denominators, units, date ranges, and uncertainty where they influence interpretation.
Find a starting skill
Clean a dataset and visualize the key trends
Task searches use OpenAI for semantic matching. Keep secrets and repository contents out of the search text. Review the suggested skill’s source and instructions before using it.
What to ask for
A profiling summary, reproducible cleaning steps, a transformed dataset, and charts with captions explaining their meaning and limitations.
Review the result
Reconcile row counts before and after cleaning. Spot-check transformed values, verify grouping and time boundaries, and distinguish associations from causal explanations.
Continue the workflow
Connect Claude Code or Codex, then copy the pinned agent prompt from a public skill page.
Related workflow: Agent skills for comparing research and tracing evidence.