Name: wandb-primary
Rating: 86.05000000000001 (1 reviews)
Author: wandb

wandb-primary

Primary W&B skill for broad or mixed Weights & Biases work: project overviews, W&B runs and artifacts, Weave traces and evaluations, Reports, and Launch workflows. Use when the task spans multiple W&B surfaces or the user asks generally what is happening in a W&B project.

2.30x

Quality

94%

Does it follow best practices?

Impact

76%

2.30x

Average score across 2 eval scenarios

Securityby

Low

Low-risk findings worth noting

Quality

Content

88%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is highly actionable and well-structured with strong validation/feedback loops and a genuine one-level reference layer, but its length and duplicated recipe blocks keep it from being fully lean. Tightening the repeated code and moving more recipe detail into references would raise conciseness and progressive disclosure.

Suggestions

De-duplicate the W&B Report authoring code (lines ~974-1010 and ~1644-1668) and the exact run-count recipe (lines ~129-140 and ~1547-1556) — keep one canonical copy and reference it.

Move more of the inline 'Fast recipes' catalog into reference docs (e.g. a RECIPES.md), leaving the body a leaner overview with pointers, to shorten the 1726-line SKILL.md.

Consolidate the repeated `sys.path.insert(0, 'skills/wandb-primary/scripts')` + import preamble into a single documented snippet referenced once, instead of restating it in every recipe.

Dimension	Reasoning	Score
Conciseness	The body assumes Claude's competence (no basic-concept padding) and is mostly dense, actionable content, but it duplicates entire recipe blocks — the W&B Report authoring code appears near-verbatim at lines ~974-1010 and ~1644-1668, and the exact run-count recipe repeats at ~129-140 and ~1547-1556 — which could be trimmed.	4 / 5
Actionability	Dozens of copy-paste-ready, executable scripts and CLI commands cover the common cases (count runs/traces/feedback, inspect runs/artifacts, eval summaries, report creation, launch submit/check), with concrete placeholders and real API calls rather than pseudocode.	5 / 5
Workflow Clarity	Multi-step flows are sequenced with explicit validation checkpoints and feedback loops — Launch (create queue → inspect → launch → check_launch), verified report save with read-back, workspace optimistic-concurrency retries, and the "CRITICAL: Large project performance rules" checklist; destructive actions are guarded ("never delete a W&B project", confirm before overwrites).	5 / 5
Progressive Disclosure	A clear routing table ("When to use what") and a "Bundled files / Reference docs" section point one level deep to 12 real reference files and helper scripts with one-line descriptions, but the 1726-line body inlines a large recipe catalog that sits between overview and reference rather than being a lean overview.	4 / 5
	Total	18 / 20 Passed

Description

100%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is third-person, concise, and simultaneously answers what the skill does and when to invoke it, enumerating concrete W&B surfaces with natural trigger phrasing. It is a strong, distinctive routing description for a primary/umbrella skill.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete capability surfaces — "project overviews, W&B runs and artifacts, Weave traces and evaluations, Reports, and Launch workflows" — giving comprehensive coverage of the skill's scope.	5 / 5
Completeness	Explicitly answers both what ("Primary W&B skill for broad or mixed ... work" enumerating every surface) and when ("Use when the task spans multiple W&B surfaces or the user asks generally what is happening in a W&B project").	5 / 5
Trigger Term Quality	Covers natural user terms and synonyms including "Weights & Biases", "W&B", "W&B runs", "Weave traces", "evaluations", "Reports", "Launch workflows", and "what is happening in a W&B project".	5 / 5
Distinctiveness Conflict Risk	"Primary W&B skill" with a clear niche (broad/mixed work spanning multiple surfaces) and distinct triggers ("spans multiple W&B surfaces", "asks generally what is happening") gives minimal conflict risk with narrower sub-skills.	5 / 5
	Total	20 / 20 Passed

Validation

93%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (1727 lines); consider splitting into references/ and linking	Warning

	Total	15 / 16 Passed

Repository: wandb/skills
Path: skills/wandb-primary/SKILL.md
Commit: b93e46c

Reviewed: 11 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.