Skill Creator
Essential for anyone building custom Claude workflows. A/B testing lets you measure whether a skill actually improves output quality vs. baseline Claude.
The meta-skill for building Claude skills. Create, test, A/B compare, and benchmark skills. Skills 2.0 features: structured evals, A/B testing, trigger-description optimizer, and retirement dates for skills that become obsolete as the base model improves.
- Build a custom automation recipe for your team's unique workflow
- Test and compare two versions of a skill to see which works better
- Benchmark your custom skill's accuracy with structured evaluations
Essential for anyone building custom Claude workflows. A/B testing lets you measure whether a skill actually improves output quality vs. baseline Claude.
Developers and teams building custom Claude Code skills. Also useful for optimizing existing skills.
https://github.com/anthropics/claude-plugins-official
By Anthropic
How to Get It
/plugin install skill-creator@claude-plugins-official
Tip: Paste this into a Claude Code conversation. Verify command matches your Claude Code version.
After installing, paste this into Claude:
Help me build a custom automation recipe for my team's unique workflow
Trust Signals ✓ Tested
Data & Access
Reviewer notes
✓ Tested review. These are observations, not a security certification.
Very powerful for power users. Tier 2 — most clients won't need this themselves, but it's valuable for consultants building client-specific workflows.
Anthropic official. Creates files but doesn't execute arbitrary code.
2026-04-02: LIMITATION: Skills under 2,000 tokens perform best per Anthropic's guide — skill-creator doesn't enforce this limit, so generated skills can creep past the threshold and degrade performance.
Things to check
- Requires setting up structured evaluation datasets upfront—skills with weak ground truth data will produce unreliable comparisons. The tool assumes you're actively monitoring skill performance; set-and-forget deployments won't benefit from retirement date features.
How to evaluate tools before deploying →
Data shown here comes from public APIs and automated scanning. Reviewer notes reflect one person's experience. This is not a security certification or legal recommendation. Always evaluate tools according to your own organization's policies.
Evaluation
Very powerful for power users. Tier 2 — most clients won't need this themselves, but it's valuable for consultants building client-specific workflows.