OpenParable
Skip to content
← Back to skills

Data Describe

Documents
data-describe

Data Describe is an Agent Skill for documents published in the dathere/qsv repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.

Generate AI-powered Data Dictionary, Description, and Tags for a CSV/TSV/Excel file

View on GitHub ↗
QUICK START

How to use this skill

Bring this guide into your coding agent with a prompt tailored to the tool you use.

  1. Open your project in Codex.
  2. Copy and paste the prompt below.
  3. Review files and risks before approving any action.
Prompt to paste
Inspect the Agent Skill at https://github.com/dathere/qsv/blob/HEAD/.claude/skills/skills/data-describe/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/data-describe, but wait for my approval before writing files or running scripts.

Copying this prompt does not install or run the skill. Review third-party files before use.

Data Describe

Generate AI-powered documentation for a tabular data file using describegpt. Produces a Data Dictionary (column labels, descriptions, types), a natural-language Description of the dataset, and semantic Tags — all via the connected LLM (no API key needed in MCP mode).

Cowork note: If relative paths don't resolve, call mcp__qsv__qsv_get_working_dir and mcp__qsv__qsv_set_working_dir to sync the working directory.

Steps

  1. Index: Run mcp__qsv__qsv_index on the file for fast random access.

  2. Profile: Run mcp__qsv__qsv_stats with cardinality: true, stats_jsonl: true to generate the stats cache. describegpt reads this cache for column metadata, so it must exist first.

  3. Describe: Run mcp__qsv__qsv_describegpt with the requested options (recommend all: true for comprehensive output). At least one inference option (dictionary, description, tags, or all) is required. Output defaults to <filestem>.describegpt.md.

  4. Present: Display the generated Data Dictionary table, Description, and Tags to the user.

Options

OptionEffect
--all (recommended)Generate Dictionary + Description + Tags in one pass
--dictionaryData Dictionary only — column labels, descriptions, types
--descriptionNatural-language dataset Description only
--tagsSemantic Tags only
--formatOutput format: Markdown (default), JSON, TSV, TOON
--languageGenerate output in a non-English language (e.g. Spanish, French)
--addl-cols-listEnrich the dictionary with extra columns (e.g. "everything", "moar!")
--tag-vocabConstrain tags to a controlled vocabulary (comma-separated)
--num-tagsNumber of tags to generate (default: 5)
--num-examplesNumber of example values per column in the dictionary
--enum-thresholdMax cardinality to treat a column as an enum in the dictionary

Notes

  • No API key needed in MCP mode — uses the connected LLM automatically via MCP sampling
  • The stats cache must exist first for best results (step 2 creates it)
  • Output defaults to <filestem>.describegpt.md
  • For Excel/JSONL files, the MCP server auto-converts to CSV first
  • Use --format JSON when you need machine-readable output for downstream processing
  • Use --language to generate documentation in the user's preferred language