Skip to main content
skill CX Codex

wandb-eval-tables — Claude Code skill: what it does, how to install

wandb-eval-tables converts W&B Table artifacts into non-destructive EvalTable previews without overwriting source runs.

Apache-2.0 393 installs ★ 76 updated 2026-10-23

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add wandb/skills --skill wandb-eval-tables -a claude-code
Open on GitHub

What wandb-eval-tables does

The skill provides a planned workflow around the table_artifact_to_eval_table.py script executed via uv. It begins with a read-only scan pass to inspect table dimensions, warnings, and eligible table keys. Once verified, it executes a single-table preview and confirms the operation using verify-preview.

All transformations produce brand-new W&B runs, EvalTables, and Weave evaluation records. The helper script relies on wandb[workspaces]>=0.28.1 alongside weave, reading credentials from WANDB_API_KEY, WANDB_ENTITY, and WANDB_PROJECT environment variables or CLI flags.

Safety boundaries stop runaway jobs: tables are capped at 10,000 rows and 100 columns, alerts fire past 1,000 rows, and runs fail if more than 50 artifacts match a single key.

What it helps with

  • inspect existing W&B Table artifacts prior to migration
  • generate non-destructive EvalTable records in separate runs
  • validate converted artifacts using verify-preview
  • clean up preview runs and Weave evaluation records

Example prompts for the agent

  • Run scan on project my-org/ml-evals using the wandb-eval-tables helper
  • Generate an EvalTable preview for the predictions table key with max-runs 4
  • Check preview integrity on run run-xyz with verify-preview

How to install wandb-eval-tables

1Skills CLI (Vercel)

npx skills add wandb/skills --skill wandb-eval-tables -a claude-code

2Manually: put the files in place

~/.claude/skills/wandb-eval-tables/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/wandb-eval-tables/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about wandb-eval-tables

How to install wandb-eval-tables in Claude Code?

Run npx skills add wandb/skills --skill wandb-eval-tables -a claude-code. The skill file will be placed at ~/.claude/skills/wandb-eval-tables/SKILL.md.

What does wandb-eval-tables do?

It guides coding agents through scanning W&B Table artifacts, converting them into typed wandb.EvalTable previews, and verifying run integrity.

What are the runtime requirements for the conversion script?

It requires uv, wandb[workspaces]>=0.28.1, weave, and a configured WANDB_API_KEY.

Source text

SKILL.md · Apache-2.0
---
name: wandb-eval-tables
description: "Convert W&B Table artifacts into non-destructive EvalTable previews with scan-first planning, typed input/output/score columns, bounded batches, verification, and safe removal. Use when a coding agent needs to create, inspect, compare, verify, or remove W&B EvalTable previews."
---
<!--
SPDX-FileCopyrightText: 2026 CoreWeave, Inc.
SPDX-License-Identifier: Apache-2.0
SPDX-PackageName: skills
-->

# W&B EvalTable previews

Use the bundled helper to preview existing `wandb.Table` data as
`wandb.EvalTable` data. A preview creates new runs and keys; it never overwrites
the source runs or tables.

Read `references/EVAL_TABLES.md` before converting. It is the canonical source
for EvalTable semantics, source selection, limits, column roles, verification,
and removal.

## Environment

Run the helper from an environment that provides a W&B SDK with
`wandb.EvalTable` and Weave. Use `uv` to supply missing dependencies:

```bash
uv run --with 'wandb[workspaces]>=0.28.1' --with weave \
  python skills/wandb-eval-tables/scripts/table_artifact_to_eval_table.py --help
```

Credentials and default scope may come from `WANDB_API_KEY`, `WANDB_ENTITY`,
and `WANDB_PROJECT`; otherwise pass explicit entity/project arguments.
Downloads use the operating system's temporary directory, not a fixed current
working directory.

## Required workflow

1. Choose the narrowest source that matches the request: exact artifact, one
   run, one sweep, or a project/workspace. Do not broaden an explicitly named
   source.
2. Run `scan`. This is read-only preflight and is not proof of conversion.
3. Review `eligible_summary_by_table_key`, table shapes, truncations, warnings,
   and existing-preview metadata. Select one table key unless the user asks for
   a broader batch.
4. Classify columns only when confident. Input tuples must uniquely identify
   rows; score columns must be numeric or boolean; free-text labels and
   rationales are outputs. Leave ambiguous tables untyped.
5. Run `preview` for exactly one `--table-key` per invocation. Keep the target
   in the source project unless the user requests another writable project.
6. If runs were created, run `verify-preview` on the final created run and the
   exact logged key. Report success only when it returns `verified: true`.
7. Report created and skipped sources plus every truncation or failed check.

```bash
T=skills/wandb-eval-tables/scripts/table_artifact_to_eval_table.py

uv run python "$T" scan \
  --source-workspace ENTITY/PROJECT --max-runs 4

uv run python "$T" preview \
  --source-workspace ENTITY/PROJECT \
  --target-project ENTITY/PROJECT \
  --max-runs 4 \
  --table-key predictions

uv run python "$T" verify-preview \
  --run ENTITY/PROJECT/PREVIEW_RUN_ID \
  --table-key predictions_preview
```

## Safety

- `scan` and `preview --dry-run` are read-only. `preview` creates W&B runs,
  EvalTables, and Weave evaluations; run it only when the user requested a
  conversion.
- The helper caps tables at 10,000 rows and 100 columns, warns above 1,000
  rows, and refuses more than 50 eligible table artifacts for one key. Surface
  every cap; never imply a truncated preview is complete.
- Do not bypass helper failures with ad-hoc logging. Narrow or correct helper
  arguments, then retry only when the change is safe.
- Removal deletes both Weave evaluations and preview runs. Run
  `delete-preview --dry-run` first, verify every target is helper-created, and
  obtain confirmation before the irreversible invocation. Never delete source
  runs.

© wandb · Apache-2.0 · wandb/skills

Similar skills

All skills for AI and ML →

From the same set5

Project sponsors

Companies that keep this analytics open