skill-creator

Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.

1.87x

Quality

88%

Does it follow best practices?

Impact

88%

1.87x

Average score across 3 eval scenarios

Securityby

Passed

No findings from the security scan

The canonical home for this skill is skill-creator in anthropics/skills

Quality

Content

77%

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

The body is a thorough, well-sequenced, highly actionable workflow with explicit feedback loops, but it carries conversational filler and redundancy that hurt token efficiency, and it points to several bundle files (the agents/ docs and eval-viewer/generate_review.py) that are absent from the actual bundle.

Suggestions

Fix or remove the broken references: the agents/grader.md, agents/comparator.md, and agents/analyzer.md files and the eval-viewer/generate_review.py path do not exist in the bundle — either add these files or correct the paths (the closest present script, scripts/generate_report.py, generates the run_loop report, not the eval viewer the workflow hinges on).

Tighten conciseness by cutting the chatty asides ("Cool? Cool.", the plumbers/grandparents paragraph, "billions a year in economic value") and stating the core loop once instead of three times, bringing the body well under its own 500-line target.

Consolidate the repeated core-loop summaries (top of file, mid-section, and the closing 'Repeating one more time' block) into a single canonical statement to remove redundancy.

Dimension	Reasoning	Score
Conciseness	The body runs ~481 lines (near its own 500-line ceiling) with conversational padding ("Cool? Cool.", the plumbers/grandparents tangent, "billions a year in economic value") and the core loop restated three times ("Repeating one more time the core loop here for emphasis"), so it is mostly efficient but could be tightened rather than fully lean.	2 / 3
Actionability	It provides concrete executable commands (e.g. "python -m scripts.aggregate_benchmark <workspace>/iteration-N --skill-name <name>"), exact JSON schemas, and precise field names ("text", "passed", "evidence"), matching the copy-paste-ready score-3 anchor; it is not vague or pseudocode-like.	3 / 3
Workflow Clarity	The multi-step process is clearly sequenced (Creating a skill sub-steps; Step 1–5 for running evals) with explicit checkpoints and a validate/review/improve/rerun feedback loop, fitting the score-3 anchor; this is not the gap-filled score-2 case.	3 / 3
Progressive Disclosure	Structure and signaling are good (one-level-deep references, a 'Reference files' section, schemas in references/schemas.md, scripts in scripts/), but several referenced bundle paths are missing — agents/grader.md, agents/comparator.md, agents/analyzer.md and the repeatedly-emphasized eval-viewer/generate_review.py do not exist in the bundle — so navigation hits dead ends and cannot reach the score-3 'easy navigation' anchor.	2 / 3
	Total	10 / 12 Passed

Description

100%

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 concise, specific, and complete: it lists several concrete actions and pairs them with an explicit 'Use when' trigger clause covering natural user phrasings. It uses third person and avoids vague fluff.

Dimension	Reasoning	Score
Specificity	"Create new skills, modify and improve existing skills, and measure skill performance" lists multiple specific concrete actions (create, modify, improve, measure, run evals, benchmark, optimize), matching the score-3 anchor rather than the partial score-2 anchor.	3 / 3
Completeness	It answers both what ("Create new skills, modify and improve existing skills, and measure skill performance") and when with an explicit "Use when..." clause, satisfying the score-3 anchor and the guideline that a missing 'Use when' caps completeness at 2.	3 / 3
Trigger Term Quality	"Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis" covers natural phrasings a user would actually say, so it sits above the score-2 'some relevant keywords but missing variations' anchor.	3 / 3
Distinctiveness Conflict Risk	The "skill"/"evals"/"benchmark"/"description triggering" niche is distinct and unlikely to fire for unrelated skills, matching the clear-niche score-3 anchor rather than the overlapping score-2 anchor.	3 / 3
	Total	12 / 12 Passed

Validation

100%

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 — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: OpenBMB/PilotDeck
Path: skills/skill-creator/SKILL.md
Commit: c88d7d1

Reviewed: about 13 hours 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.