create-skill
Autonomously build, test, and validate a new Claude Code skill from a natural language description.
Skill by jasonbhorne
Chalk Labs
Need help deploying create-skill?
Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/jasonbhorne-create-skill
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: testing, agent-building
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/create-skill
What's inside
- Skill name
create-skill- Description
- Autonomously build, test, and validate a new Claude Code skill from a natural language description.
- License
- MIT
- Instructions
- 333 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
Build a new Claude Code skill from the user's description. Handles the full lifecycle: scaffold, test, fix, validate, report.
Skill description: $ARGUMENTS
## Overview
You are a meta-skill that autonomously creates new Claude Code skills. You will:
1. Parse the user's description to determine skill type and requirements
2. Create the skill file with proper structure
3. Generate a pytest test suite
4. Run tests, fix failures (up to 3 iterations)
5. Validate the skill is loadable
6. Generate a validation report
7. Present the finished skill only after all checks pass
## Step 1: Parse the Description
Analyze `$ARGUMENTS` to determine:
- **Skill name**: derive a short kebab-case slug (e.g., "budget-report", "meeting-notes")
- **Skill type**: determine if this should be a full skill (`~/.claude/skills/<name>/SKILL.md`) or a simple command (`~/.claude/commands/<name>.md`). Use a skill for anything with embedded Python, agent teams, or complex logic. Use a command for simple prompt-based workflows (like research templates).
- **Output format**: what does the skill produce? (.docx, .xlsx, .pptx, .md, .pdf, .csv, or text output)
- **Key behaviors**: what specific actions, transformations, or analyses does it perform?
- **Input requirements**: what arguments does it need?
- **Edge cases**: empty input, missing directories, missing files, invalid arguments
If the description is too vague to build a working skill, use AskUserQuestion to clarify before proceeding. Otherwise, make reasonable assumptions and note them.
## Step 2: Create the Team
```
TeamCreate with team_name: "create-skill-<name>"
```
Create 4 tasks:
1. **Build the skill file** (builder agent)
2. **Write the test suite** (tester agent)
3. **Run tests and iterate** (coordinator — you)
4. **Generate validation report** (coordinator — you)
Set task 3 as blocked by tasks 1 and 2. Set task 4 as blocked by task 3.
## Step 3: Spawn the Builder Agent
Spawn a `general-purpose` agent named `builder` with `team_name` set. Give it these instructions:
### Builder Agent Instructions
Create a Claude Code skill at the determined path. Follow these patterns exactly:
**For skills (`~/.claude/skills/<name>/SKILL.md`):**
```yaml
---
name: <slug>
description: <one-line description>
argument-hint: "<usage hint>"
---
```
Followed by the full implementation. Reference these conventions from existing skills:
- Use `$ARGUMENTS` for the full argument string
- For Python-based skills: embed the complete Python code in the SKILL.md as fenced code blocks, with instructions to write to `/tmp/` and execute via Bash
- For document-producing skills: include the full python-docx/openpyxl/python-pptx code inline
- Steps should be numbered and explicit
- Include a "Rules" section at the end with constraints
- Author metadata: "$AUTHOR_NAME" where applicable
- Output paths: `~/Documents/` or `~/Downloads/` depending on type
- Always clean up temp files after execution
- Always open the output file with `open` command at the end
- Always print a summary report of what was created
**For commands (`~/.claude/commands/<name>.md`):**
```markdown
# Title: $ARGUMENTS
Instructions here...
```
No YAML frontmatter. Plain markdown with instructions.
**Critical rules for the builder:**
- Study the existing skills at `~/.claude/skills/` to match the style and conventions
- If the skill produces .docx files, use the python-docx + lxml comment insertion pattern from doc-review, NOT raw zipfile manipulation (unless inserting comments into existing docs, which needs the 3-file XML approach for Word 2016+ compatibility)
- If the skill produces .xlsx files, use openpyxl with proper styling (reference time-savings skill)
- If the skill produces .pptx files, use the full helper function set from the pptx skill
- Make the skill fully self-contained — no imports from custom modules
- Test the embedded Python by running it mentally for syntax correctness
- Include error handling for: file not found, missing directories (create them), invalid input
After creating the file, send the file path and a brief summary back to the coordinator.
## Step 4: Spawn the Tester Agent
Spawn a `general-purpose` agent named `tester` with `team_name` set. Give it these instructions:
### Tester Agent Instructions
Create a pytest test suite at `/tmp/test_skill_<name>.py` that validates the skill. The test suite must cover:
**1. File creation test:**
```python
def test_skill_file_exists():
"""Verify the skill file was created at the correct path."""
assert os.path.exists(SKILL_PATH)
```
**2. Syntax validation test:**
```python
def test_skill_syntax():
"""Verify YAML frontmatter parses correctly (for skills) or markdown is valid (for commands)."""
```
- For skills: parse the YAML frontmatter, verify `name` and `description` fields exist
- For commands: verify the file starts with `# ` and contains `$ARGUMENTS`
**3. Output format tests (based on what the skill produces):**From jasonbhorne/claude-code-skills, MIT.
More from jasonbhorne/claude-code-skills
- 0.474score
image-prep
jasonbhorne
Batch-prep photos for communications (newsletters, decks, websites).
SkillCFreshness: CClaude Code +9 - 0.474score
job-finder
jasonbhorne
Find and rank current job openings for a friend or candidate.
SkillCFreshness: CClaude Code +9 - 0.471score
fantasy-football
jasonbhorne
AI-powered fantasy football assistant with draft prep, live draft, weekly matchup, and trade analysis modes.
SkillCFreshness: CClaude Code +9 - 0.471score
financial-advisor
jasonbhorne
Monthly personal financial analysis with XLSX dashboard and DOCX narrative report.
SkillCFreshness: CClaude Code +9 - 0.469score
blog
jasonbhorne
Write a short, punchy blog post for $AUTHOR_WEBSITE in the user's voice.
SkillCFreshness: CClaude Code +9 - 0.406score
notebooklm-pipeline
jasonbhorne
Automated NotebookLM pipeline: create notebook, add sources, then generate video podcast + report + infographic + slides in parallel.
SkillDFreshness: DClaude Code +9 - 0.406score
ai-paper-review
jasonbhorne
Review a student's "AI Leading and Learning" paper and insert Word comments with rubric-based feedback.
SkillDFreshness: DClaude Code +9 - 0.406score
dissertator
jasonbhorne
Review an university EDLD dissertation and insert Word comments with rigorous, constructive feedback.
SkillDFreshness: DClaude Code +9 - 0.406score
doc-review
jasonbhorne
Review a .docx file and insert Word comments for typos, errors, formatting issues, and other feedback.
SkillDFreshness: DClaude Code +9 - 0.406score
fact-check
jasonbhorne
Fact-check a .docx file for internal consistency, unsourced claims, citation integrity, logical coherence, plausibility issues, and stale references.
SkillDFreshness: DClaude Code +9 - 0.404score
check-in
jasonbhorne
Clean up Friday leadership check-in notes into polished DOCX/PDF files and maintain an action item tracker.
SkillDFreshness: DClaude Code +9 - 0.404score
newsletter-link
jasonbhorne
Add a link to the must-include list for the next AI newsletter.
SkillDFreshness: DClaude Code +9
Signals
- Stars
- 0
- Forks
- 2
- Open issues
- 0
- Last commit (this folder)
- Mar 7, 2026 (7 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from jasonbhorne
- 0.474score
image-prep
jasonbhorne
Batch-prep photos for communications (newsletters, decks, websites).
SkillCFreshness: CClaude Code +9 - 0.474score
job-finder
jasonbhorne
Find and rank current job openings for a friend or candidate.
SkillCFreshness: CClaude Code +9 - 0.471score
financial-advisor
jasonbhorne
Monthly personal financial analysis with XLSX dashboard and DOCX narrative report.
SkillCFreshness: CClaude Code +9 - 0.471score
fantasy-football
jasonbhorne
AI-powered fantasy football assistant with draft prep, live draft, weekly matchup, and trade analysis modes.
SkillCFreshness: CClaude Code +9 - 0.469score
blog
jasonbhorne
Write a short, punchy blog post for $AUTHOR_WEBSITE in the user's voice.
SkillCFreshness: CClaude Code +9 - 0.406score
fact-check
jasonbhorne
Fact-check a .docx file for internal consistency, unsourced claims, citation integrity, logical coherence, plausibility issues, and stale references.
SkillDFreshness: DClaude Code +9
Done for you
Rather not set it up yourself?
Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.
See how we build AI agents →Read the case studies
Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.
- launches shipped
- 50+
- launches shipped
- media placements
- 500+
- media placements
- founders on every account
- 2
- founders on every account
Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block
Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.
Get create-skill running in your business
Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.
Cite this page
create-skill. Agent Mart. https://agentmart.dev/skills/jasonbhorne-create-skill (last updated 2026-10-24)