# Skill Creator

> Writes a new skill or improves one with evals, a benchmark, graded versions and description tuning. Use when adding a new skill to the catalog, turning a workflow into a skill, comparing runs with and without a skill, or when a skill never gets picked up.

- Skill: `niksavis/skill-creator` (Agent Skill, multi-file: 18 files)
- Install (CLI): `npx get-skill add niksavis/skill-creator`
- Raw SKILL.md: https://api.skillmd.com/api/skills/niksavis/skill-creator/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Productivity
- License: Apache-2.0
- Author: niksavis (https://skillmd.com/u/niksavis)
- Updated: 2026-11-11
- Page: https://skillmd.com/skills/niksavis/skill-creator

---

<!-- Generated by `basicly skills-build` from skill.yaml. Do not edit; edit the source. -->

# Skill Creator

> Modified by basicly on 2026-10-02 from anthropics/skills skill-creator at 8a1541c4 (Apache-2.0, see `LICENSE.txt`).

Create a skill, test it on real prompts, measure it, and improve it until the user is
satisfied. This skill owns the eval loop.

The loop:

1. Decide what the skill does and roughly how.
2. Write a draft as a `skill.yaml` source.
3. Write the trigger cases and task cases in `evals.yaml`, then run a few task cases with and without the skill.
4. Evaluate the results with the user, both by reading them and by the numbers.
5. Rewrite the skill from the feedback and from flaws that the benchmark shows.
6. Repeat. Then add task cases and run again at a larger scale.

Find out where the user is in this loop and continue from there. A user with a draft goes
straight to the eval step. A user who wants no formal evals gets a quick draft and a
conversation. Description tuning comes last.

## Capture the intent

The conversation may already hold the workflow, for example "turn this into a skill".
Then take the answers from it first: the tools used, the order of steps, the corrections
the user made, the input and output formats. Ask only for missing facts that change the result.

1. What does the skill let the agent do?
2. When does it trigger: which phrases and situations?
3. What is the expected output?
4. Does it need task cases? A skill with a verifiable output (a file transform, data
   extraction, generated code, fixed workflow steps) benefits from them. A skill with a
   subjective output (writing style, design) often does not. Suggest a default and let the
   user decide.

Ask about edge cases, formats, example files, success criteria and dependencies before
you write task cases. Before you build, check that no installed or published skill
already does the job: the `find-skills` skill does that search.

## Read when needed

- Before source authoring, read [authoring](references/authoring.md); `catalog-authoring` owns catalog format and projections.
- For trigger cases, test runs with and without the skill, grading and timing, read [evaluation](references/evaluation.md). The run with the skill and the run without it may go to parallel agents; you write the skill yourself.
- For feedback, blind comparison, description tuning or evaluator script commands, read [improvement](references/improvement.md). Execute the scripts; do not paste their source into the prompt.
- Read [grader](references/grader.md), [comparator](references/comparator.md), [analyzer](references/analyzer.md) or [schemas](references/schemas.md) only for the step that names it.
- Done when every task case passes with the skill, graded by an agent that did not write it, every trigger case passes in `basicly catalog lint`, and `basicly skills-check` passes.

