skill collection · Study the method
skill-creator
Create or refine a skill and study how evaluations are organized.
anthropics/skillsskills/skill-creator/SKILL.md
Before you use it
- Input
- A task description, an existing skill or examples of desired behavior.
- Expected output
- Skill instructions and an evaluation/improvement workflow.
- Prerequisites
- Includes scripts, evaluation resources and references. Inspect them before use; observed structure is not an execution audit.
Publisher documentation for this source collection. Review the chosen skill and client setup; no runtime observation is recorded.
Publisher documentation, static file observations and environment tests are different kinds of evidence. Runtime behavior has not been observed.
Study the method
Trace instructions → evaluation inputs → graders → review output rather than copying only the frontmatter.
18 files in the observed scope; 9 in scripts/, 1 in references/, 1 in assets/.
- agents/analyzer.md ↗
- agents/comparator.md ↗
- agents/grader.md ↗
- assets/eval_review.html ↗
- eval-viewer/generate_review.py ↗
- eval-viewer/viewer.html ↗
- LICENSE.txt ↗
- references/schemas.md ↗
- scripts/__init__.py ↗
- scripts/aggregate_benchmark.py ↗
This is a sample of observed paths, not the full directory. Zero files in scripts/ does not mean no execution, network access or file writes. Markdown and other folders may describe or contain executable behavior.
Source version and requirements
- Observed
- 2026-09-07T13:56:32.102Z
- Source commit
- 41bbe19d1a1a7eaab5e7bb9050a417e5c6cffc8f
- Content SHA-256
- dcd4803e61e913e6fc27294184cd3a71f09f5e924ff20c8a9a20173e7b3c2bcf
- License
- Apache-2.0
- Author attribution
- Repository owner: anthropics. Individual authorship is not independently established.
This page describes a pinned source snapshot, not the latest repository state. Changes after this commit have not been reconciled here. A repository license does not necessarily cover external models, datasets or services.
Open current repository ↗