arize-dataset

Creates, manages, and queries Arize datasets and examples. Covers dataset CRUD, appending examples, exporting data, and file-based dataset creation using the ax CLI. Use when the user needs test data, evaluation examples, or mentions create dataset, list datasets, export dataset, append examples, dataset version, golden dataset, or test set.

3,107
installs: 0 up / 0 down, score 0.00
votes: 65
repo stars
Updated: less than a month ago
In the catalog since: 1 Oct 2026
From: arize-ai/arize-skills

Get it through your agent npx skills add arize-ai/arize-skills@arize-dataset

Skill instructions

Rendered from the published SKILL.md, 335 lines.

Arize Dataset Skill

SPACE — --space flags accept a space name (e.g., my-workspace) or a base64 space ID (e.g., U3BhY2U6...). Find yours with ax spaces list.

Concepts

Dataset = a versioned collection of examples used for evaluation and experimentation
Dataset Version = a snapshot of a dataset at a point in time; updates can be in-place or create a new version
Example = a single record in a dataset with arbitrary user-defined fields (e.g., question, answer, context)
Space = an organizational container; datasets belong to a space

System-managed fields on examples (id, created_at, updated_at) are auto-generated by the server -- never include them in create or append payloads.

Prerequisites

Proceed directly with the task — run the ax command you need. Do NOT check versions, env vars, or profiles upfront.

If an ax command fails, troubleshoot based on the error:

command not found or version error → see references/ax-setup.md
401 Unauthorized / missing API key → run ax profiles show to inspect the current profile. If the profile is missing or the API key is wrong, follow references/ax-profiles.md to create/update it. If the user doesn't have their key, direct them to https://app.arize.com/admin > API Keys
Space unknown → see Space
Project unclear → ask the user, or run ax projects list -o json --limit 100 and present as selectable options
Security: Never read .env files or search the filesystem for credentials. Exception: the non-secret ARIZE_SPACE_ID line (see Space). Use ax profiles for Arize credentials and ax ai-integrations for LLM provider keys. Never ask the user to paste secrets into chat. For missing credentials, see references/ax-profiles.md.

List Datasets: `ax datasets list`

Browse datasets in a space. Listing is for human browsing; when the dataset name or ID is known, call ax datasets get NAME_OR_ID --space SPACE -o json directly.

ax datasets list
ax datasets list --space SPACE --limit 20
ax datasets list --cursor CURSOR_TOKEN
ax datasets list -o json

For all flags, run ax datasets list --help.

Get Dataset: `ax datasets get`

Quick metadata lookup -- returns dataset name, space, timestamps, and version list.

ax datasets get NAME_OR_ID
ax datasets get NAME_OR_ID -o json
ax datasets get NAME_OR_ID --space SPACE   # required when using dataset name instead of ID

For all flags, run ax datasets get --help.

Response fields

Field	Type	Description
`id`	string	Dataset ID
`name`	string	Dataset name
`space_id`	string	Space this dataset belongs to
`created_at`	datetime	When the dataset was created
`updated_at`	datetime	Last modification time
`versions`	array	List of dataset versions (id, name, dataset_id, created_at, updated_at)

Export Dataset: `ax datasets export`

Download examples to a file. By default, export returns one page of examples (--limit sets the page size, --cursor selects a later page). Pass --all to export the full dataset. Use --filter for server-side narrowing with a SQL-like expression over example columns, id, and annotation.<name>.*; --filter is not supported with --all.

ax datasets export NAME_OR_ID
# -> dataset_abc123_20260305_141500/examples.json

ax datasets export NAME_OR_ID --all
ax datasets export NAME_OR_ID --filter "annotation.Correctness.label = 'Correct'"
ax datasets export NAME_OR_ID --version-id VERSION_ID
ax datasets export NAME_OR_ID --output-dir ./data
ax datasets export NAME_OR_ID --stdout
ax datasets export NAME_OR_ID --stdout | jq '.[0]'
ax datasets export NAME_OR_ID --space SPACE   # required when using dataset name instead of ID

For all flags, run ax datasets export --help.

Agent rule: Pass --all whenever the task needs every example; an export without --all stops after one page.

Export completeness verification: After exporting, confirm the row count matches what the server reports:

# Get the server-reported count from dataset metadata
ax datasets get DATASET_NAME --space SPACE -o json | jq '.versions[-1] | {version: .id, examples: .example_count}'

# Compare to what was exported
jq 'length' dataset_*/examples.json

# If counts differ, re-export with --all

Output is a JSON array of example objects. Each example has system fields (id, created_at, updated_at) plus all user-defined fields:

[
  {
    "id": "ex_001",
    "created_at": "2026-01-15T10:00:00Z",
    "updated_at": "2026-01-15T10:00:00Z",
    "question": "What is 2+2?",
    "answer": "4",
    "topic": "math"
  }
]

Create Dataset: `ax datasets create`

Create a new dataset from a data file (--file) or inline JSON (--json).

ax datasets create --name "My Dataset" --space SPACE --file data.csv
ax datasets create --name "My Dataset" --space SPACE --file data.json
ax datasets create --name "My Dataset" --space SPACE --file data.jsonl
ax datasets create --name "My Dataset" --space SPACE --file data.parquet
ax datasets create --name "My Dataset" --space SPACE --json '[{"question": "What is 2+2?", "answer": "4"}]'

For all flags, run ax datasets create --help.

Show the whole skill (215 more lines)

Passing data via stdin

Use --file - to pipe data from another command:

echo '[{"question": "What is 2+2?", "answer": "4"}]' | ax datasets create --name "my-dataset" --space SPACE --file -

# Or with a heredoc
ax datasets create --name "my-dataset" --space SPACE --file - << 'EOF'
[{"question": "What is 2+2?", "answer": "4"}]
EOF

Supported file formats

Format	Extension	Notes
CSV	`.csv`	Column headers become field names
JSON	`.json`	Array of objects
JSON Lines	`.jsonl`	One object per line (NOT a JSON array)
Parquet	`.parquet`	Column names become field names; preserves types

Format gotchas:

CSV: Loses type information — dates become strings, null becomes empty string. Use JSON/Parquet to preserve types.
JSONL: Each line is a separate JSON object. A JSON array ([{...}, {...}]) in a .jsonl file will fail — use .json extension instead.
Parquet: Preserves column types. Requires pandas/pyarrow to read locally: pd.read_parquet("examples.parquet").

Append Examples: `ax datasets append`

Add examples to an existing dataset. Two input modes -- use whichever fits.

Inline JSON (agent-friendly)

Generate the payload directly -- no temp files needed:

ax datasets append DATASET_NAME --space SPACE --json '[{"question": "What is 2+2?", "answer": "4"}]'

ax datasets append DATASET_NAME --space SPACE --json '[
  {"question": "What is gravity?", "answer": "A fundamental force..."},
  {"question": "What is light?", "answer": "Electromagnetic radiation..."}
]'

From a file

ax datasets append DATASET_NAME --space SPACE --file new_examples.csv
ax datasets append DATASET_NAME --space SPACE --file additions.json

To a specific version

ax datasets append DATASET_NAME --space SPACE --json '[{"q": "..."}]' --version-id VERSION_ID

For all flags, run ax datasets append --help.

Exactly one of --json or --file is required.

Validation

Each example must be a JSON object with at least one user-defined field
Maximum 100,000 examples per request

Schema validation before append: If the dataset already has examples, inspect its schema before appending to avoid silent field mismatches:

# Check existing field names in the dataset
ax datasets export DATASET_NAME --space SPACE --stdout | jq '.[0] | keys'

# Verify your new data has matching field names
echo '[{"question": "..."}]' | jq '.[0] | keys'

# Both outputs should show the same user-defined fields

Fields are free-form: extra fields in new examples are added, and missing fields become null. However, typos in field names (e.g., queston vs question) create new columns silently -- verify spelling before appending.

Update Examples: `ax datasets update-examples`

ax datasets update-examples patches rows by example id; ax datasets delete-examples removes rows from one version and requires --version-id plus --example-ids as a JSON array of example IDs. Uncommon — run ax datasets update-examples --help or ax datasets delete-examples --help for flags and input formats.

Delete Dataset: `ax datasets delete`

ax datasets delete NAME_OR_ID
ax datasets delete NAME_OR_ID --space SPACE   # required when using dataset name instead of ID
ax datasets delete NAME_OR_ID --force   # skip confirmation prompt

For all flags, run ax datasets delete --help.

Update Dataset: `ax datasets update`

Rename an existing dataset.

ax datasets update NAME_OR_ID --name "new-dataset-name"
ax datasets update NAME_OR_ID --name "new-dataset-name" --space SPACE

For all flags, run ax datasets update --help.

Annotate Examples: `ax datasets annotate-examples`

Write annotations onto dataset examples in bulk from a file. Upsert semantics — existing annotations with the same key are updated, new ones are created. Up to 1000 examples per request. Each record needs a record_id (the dataset example ID) and values; unmatched record IDs are silently ignored, so confirm IDs against an export first.

ax datasets annotate-examples NAME_OR_ID --file annotations.json
ax datasets annotate-examples NAME_OR_ID --file annotations.csv --space SPACE

For all flags, run ax datasets annotate-examples --help.

Workflows

Find a dataset by name

All dataset commands accept a name or ID directly (add --space SPACE when passing a name). To resolve a name to its ID, read it from get:

ax datasets get "eval-set-v1" --space SPACE -o json | jq -r '.id'

Create a dataset from file for evaluation

Prepare a CSV/JSON/Parquet file with your evaluation columns (e.g., input, expected_output)
- If generating data inline, pass it with --json or pipe it via --file - (see the Create Dataset section)
ax datasets create --name "eval-set-v1" --space SPACE --file eval_data.csv
Verify: ax datasets get DATASET_NAME --space SPACE
Use the dataset name to run experiments

Add examples to an existing dataset

# Append inline or from a file using the dataset name (see Append Examples section for full syntax)
ax datasets append DATASET_NAME --space SPACE --json '[{"question": "...", "answer": "..."}]'
ax datasets append DATASET_NAME --space SPACE --file additional_examples.csv

Download dataset for offline analysis

ax datasets export DATASET_NAME --space SPACE --all -- download every example to file
Parse the JSON: jq '.[] | .question' dataset_*/examples.json

Export a specific version

# List versions
ax datasets get DATASET_NAME --space SPACE -o json | jq '.versions'

# Export that version
ax datasets export DATASET_NAME --space SPACE --version-id VERSION_ID

Iterate on a dataset

Export current version: ax datasets export DATASET_NAME --space SPACE --all
Modify the examples locally
Append new rows: ax datasets append DATASET_NAME --space SPACE --file new_rows.csv
Or create a fresh version: ax datasets create --name "eval-set-v2" --space SPACE --file updated_data.json

Pipe export to other tools

# Count examples
ax datasets export DATASET_NAME --space SPACE --all --stdout | jq 'length'

# Extract a single field
ax datasets export DATASET_NAME --space SPACE --stdout | jq '.[].question'

# Convert to CSV with jq
ax datasets export DATASET_NAME --space SPACE --stdout | jq -r '.[] | [.question, .answer] | @csv'

Dataset Example Schema

Examples are free-form JSON objects. There is no fixed schema -- columns are whatever fields you provide. System-managed fields are added by the server:

Field	Type	Managed by	Notes
`id`	string	server	Auto-generated UUID. Required on update, forbidden on create/append
`created_at`	datetime	server	Immutable creation timestamp
`updated_at`	datetime	server	Auto-updated on modification
(any user field)	any JSON type	user	String, number, boolean, null, nested object, array

Related Skills

arize-trace: Export production spans to understand what data to put in datasets → use arize-trace
arize-experiment: Run evaluations against this dataset → next step is arize-experiment
arize-prompt-optimization: Use dataset + experiment results to improve prompts → use arize-prompt-optimization

Troubleshooting

Problem	Solution
`ax: command not found`	See references/ax-setup.md
`401 Unauthorized`	API key is wrong, expired, or doesn't have access to this space. Fix the profile using references/ax-profiles.md.
`No profile found`	No profile is configured. See references/ax-profiles.md to create one.
`Dataset not found`	Check the name or ID with `ax datasets get NAME_OR_ID --space SPACE`; use `ax datasets list --space SPACE` only to browse for the right name
`File format error`	Supported: CSV, JSON, JSONL, Parquet. Use `--file -` to read from stdin.
`platform-managed column`	Remove `id`, `created_at`, `updated_at` from create/append payloads
`reserved column`	Remove `time`, `count`, or any `source_record_*` field
`Provide either --json or --file`	Append requires exactly one input source
`Examples array is empty`	Ensure your JSON array or file contains at least one example
`not a JSON object`	Each element in the `--json` array must be a `{...}` object, not a string or number

Save Credentials for Future Use

See references/ax-profiles.md § Save Credentials for Future Use.

Get it through your agent

Call the Graph-of-Skills MCP server's retrieve_skill_bundle tool and it returns this skill's text inside your budget.

{
  "tool": "retrieve_skill_bundle",
  "arguments": {
    "query": "arize-dataset"
  }
}

New to this? Set up Graph-of-Skills or get a free key. The text on this page comes from its authors: how skills are sourced and what we check. If none fits your task, your agent can write a skill for your own library. To keep a skill from your own library on a machine, see how to install one of your skills.

Share on X Share on LinkedIn Share on Bluesky Share on Hacker News

Share on Reddit

Source

Source repository arize-ai/arize-skills
Listing on skills.sh

From arize-ai/arize-skills via skills.sh

Related skills

arize-experiment
Creates, runs, and analyzes Arize experiments for evaluating and comparing model performance. Covers experiment CRUD, exporting runs, comparing results, and evaluation workflows using the ax CLI. Use when the user mentions create experiment, run experiment, compare models, model performance, evaluate AI, experiment results, benchmark, A/B test models, or measure accuracy.
3,109 installs
langsmith-dataset
INVOKE THIS SKILL when creating evaluation datasets, uploading datasets to LangSmith, or managing existing datasets. Covers dataset types (final_response, single_step, trajectory, RAG), CLI management commands, SDK-based creation, and example management. Uses the langsmith CLI tool.
4,789 installs
arize-prompt-optimization
Optimizes, improves, and debugs LLM prompts using production trace data, evaluations, and annotations. Extracts prompts from spans, gathers performance signal, and runs a data-driven optimization loop using the ax CLI. Use when the user mentions optimize prompt, improve prompt, make AI respond better, improve output quality, prompt engineering, prompt tuning, or system prompt improvement.
3,280 installs
google-analytics-data-api-basics
Manages Google Analytics reporting data, enables the Analytics Data API via the Cloud CLI, and creates reports using the Google Analytics Data API (v1beta). Use when you need to interact with Google Analytics properties, run customized analytics reports, query metrics (like activeUsers, screenPageViews) and dimensions (like city, date), check metrics and dimensions compatibility, or verify API enablement. Don't use for Google Analytics Admin API operations (e.g., creating properties, managing users) or for front-end tracking installation.
7,389 installs
explore-data
Profile and explore a dataset to understand its shape, quality, and patterns. Use when encountering a new table or file, checking null rates and column distributions, spotting data quality issues like duplicates or suspicious values, or deciding which dimensions and metrics to analyze.
5,834 installs

More from arize-ai/arize-skills

arize-instrumentation
Adds Arize AX tracing to an LLM application for the first time. Detects the stack, routes to the single matching integration doc, wires auto-instrumentation after user confirmation, and verifies traces land. Use when the user wants to instrument their app, add tracing from scratch, set up LLM observability, integrate OpenTelemetry or openinference, or get started with Arize tracing.
3,440 installs
arize-prompt-optimization
Optimizes, improves, and debugs LLM prompts using production trace data, evaluations, and annotations. Extracts prompts from spans, gathers performance signal, and runs a data-driven optimization loop using the ax CLI. Use when the user mentions optimize prompt, improve prompt, make AI respond better, improve output quality, prompt engineering, prompt tuning, or system prompt improvement.
3,280 installs
arize-trace
Downloads, exports, and inspects existing Arize traces and spans to understand what an LLM app is doing or debug runtime issues. Covers exporting traces by ID, spans by ID, sessions by ID, and root-cause investigation using the ax CLI. Use when the user wants to look at existing trace data, see what their LLM app is doing, export traces, download spans, investigate errors, or analyze behavior regressions.
3,223 installs
arize-experiment
Creates, runs, and analyzes Arize experiments for evaluating and comparing model performance. Covers experiment CRUD, exporting runs, comparing results, and evaluation workflows using the ax CLI. Use when the user mentions create experiment, run experiment, compare models, model performance, evaluate AI, experiment results, benchmark, A/B test models, or measure accuracy.
3,109 installs
arize-evaluator
Handles LLM-as-judge and code evaluator workflows on Arize including creating/updating evaluators, running evaluations on spans or experiments, managing tasks, trigger-run operations, column mapping, and continuous monitoring. Use when the user mentions create evaluator, LLM judge, code evaluator, hallucination, faithfulness, correctness, relevance, run eval, score spans, score experiment, trigger-run, column mapping, continuous monitoring, or improve evaluator prompt.
3,079 installs

Content is provided as published by its creators; you are responsible for how you use it.

To request removal, use the contact form.