Assistant Skills

10x

_{Faster skill
development}

6

_{Production-ready
skills included}

80+

_{Scripts &
templates}

0

_{Boilerplate
to write}

Build Claude Code skills in minutes, not days.
_{Templates, wizards, and tools from production implementations.}

> "Create a new Datadog Assistant Skills project"

✓ Project scaffolded with best practices
✓ Shared library configured
✓ First skill template ready
✓ Test infrastructure in place

Get Started • Skills • Use Cases • Templates

The Difference

Starting from Scratch

1. Research skill structure
2. Figure out SKILL.md format
3. Write boilerplate code
4. Set up test infrastructure
5. Create documentation
6. Debug token efficiency issues

Hours of trial and error...

With Assistant Skills

"Create a new GitHub Assistant Skills project
with search and issues skills"

Production-ready in minutes.

Time Saved

Task	From Scratch	With This Toolkit	Saved
New project setup	2-4 hours	5 minutes	95%
Add new skill	30-60 min	2 minutes	95%
Optimize for tokens	1-2 hours	1 minute	98%
Create landing page	2-3 hours	10 minutes	90%

Quick Start

1. Install Plugin

From Claude Code:

/plugin grandcamel/Assistant-Skills

This adds the marketplace and installs the assistant-skills plugin with all 6 skills.

Alternative: Clone locally

git clone https://github.com/grandcamel/Assistant-Skills.git
cd Assistant-Skills && /plugin .

2. Set Up Environment

Run the setup wizard to configure the shared Python environment:

/assistant-skills-setup

The wizard will:

Create a shared venv at ~/.assistant-skills-venv/
Install Python dependencies
Add the claude-as shell function to your .bashrc/.zshrc

After setup, use claude-as instead of claude to run with dependencies:

claude-as  # Runs Claude with Assistant Skills venv

Alternative: Docker

Run Claude Code with Assistant Skills in a container—no local Python setup required:

# Using the helper script
./scripts/claude-as-docker.sh

# Or directly with Docker
docker run -it --rm \
  -e ANTHROPIC_API_KEY=$ANTHROPIC_API_KEY \
  -v $(pwd):/workspace \
  -v ~/.claude:/home/claude/.claude \
  ghcr.io/grandcamel/assistant-skills:latest

Install additional plugins or marketplaces:

# Install extra plugins
CLAUDE_PLUGINS="owner/plugin1,owner/plugin2" ./scripts/claude-as-docker.sh

# Install from multiple marketplaces
CLAUDE_MARKETPLACES="grandcamel/Assistant-Skills,other/marketplace" ./scripts/claude-as-docker.sh

See Docker Usage for more options.

3. Create Your First Project

"Create a new Slack Assistant Skills project"

Or use the interactive wizard:

/assistant-builder-setup

4. Start Building

Claude scaffolds your project with:

Optimized directory structure
Shared library with HTTP client, error handling
First skill template ready to customize
Test infrastructure configured

That's it. Start building your skills immediately.

Included Skills

Skill	Purpose	Example
setup-wizard	Configure shared venv & shell	`/assistant-skills-setup`
assistant-builder	Create & extend projects	`"Add a search skill to my project"`
skills-optimizer	Audit token efficiency	`"Analyze my skill for optimization"`
landing-page	Generate branded READMEs	`"Create a landing page for this project"`
library-publisher	Publish shared libs to PyPI	`"Publish my shared library as a PyPI package"`
e2e-testing	Set up E2E test infrastructure	`"Add E2E tests to my plugin"`

setup-wizard

Configure the shared Python environment for all Assistant Skills plugins.

# Run the interactive setup wizard
/assistant-skills-setup

# After setup, use claude-as to run with dependencies
claude-as

Creates ~/.assistant-skills-venv/ (shared venv) and adds claude-as function to shell.

assistant-builder

Interactive wizard for creating new Assistant Skills projects or adding skills to existing ones.

# Create new project
python skills/assistant-builder/scripts/scaffold_project.py

# Add skill to existing project
python skills/assistant-builder/scripts/add_skill.py --name "search"

# Validate project structure
python skills/assistant-builder/scripts/validate_project.py /path/to/project

skills-optimizer

Audit skills for token efficiency and progressive disclosure compliance.

# Analyze a skill (get grade A-F)
./skills/skills-optimizer/scripts/analyze-skill.sh ~/.claude/skills/my-skill

# Audit all skills
./skills/skills-optimizer/scripts/audit-all-skills.sh ~/.claude/skills

landing-page

Generate professional README landing pages with consistent branding.

# Analyze project metadata
python skills/landing-page/scripts/analyze_project.py /path/to/project

# Generate logo SVG
python skills/landing-page/scripts/generate_logo.py --name jira --primary "#0052CC"

library-publisher

Extract shared libraries and publish them as PyPI packages with automated CI/CD.

# Analyze existing shared library
python skills/library-publisher/scripts/analyze_library.py /path/to/project

# Scaffold PyPI package
python skills/library-publisher/scripts/scaffold_package.py \
  --name "myproject-lib" \
  --source /path/to/lib \
  --output ~/myproject-lib

# Migrate project to use new package
python skills/library-publisher/scripts/migrate_imports.py \
  --project /path/to/project \
  --package myproject_lib

e2e-testing

Set up end-to-end testing infrastructure for Claude Code plugins.

# Initialize E2E testing infrastructure
python skills/e2e-testing/scripts/setup_e2e.py /path/to/project

# Auto-generate test cases from plugin structure
python skills/e2e-testing/scripts/generate_test_cases.py /path/to/project

# Run tests
./scripts/run-e2e-tests.sh

# Update documentation with E2E info
python skills/e2e-testing/scripts/update_docs.py /path/to/project

Who Is This For?

Developers building Claude Code integrations

Skip the boilerplate—start with production patterns
Follow proven architecture from 40+ skill implementations
Get token-efficient skills that don't bloat context
Use TDD workflow with test templates included

Teams standardizing on Claude Code

Consistent skill structure across projects
Shared library patterns for common functionality
Documentation templates for team onboarding
Quality gates with optimization scoring

Open source maintainers

Professional landing pages in minutes
Branded logos with terminal prompt design
Consistent visual identity across repos
Badge and stats automation

Templates

Comprehensive templates derived from production implementations:

Folder	Purpose
`templates/00-project-lifecycle/`	API research, GAP analysis, architecture planning
`templates/01-project-scaffolding/`	Project initialization, directory structure, configs
`templates/02-shared-library/`	HTTP client, error handling, auth patterns
`templates/03-skill-templates/`	SKILL.md format, script templates, validators
`templates/04-testing/`	TDD workflow, pytest fixtures, test patterns
`templates/05-documentation/`	Workflow guides, reference docs, examples
`templates/06-git-and-ci/`	Commit conventions, GitHub Actions, releases

Shared Library

Production-ready Python utilities available via PyPI:

pip install assistant-skills-lib

Module	Purpose
`formatters`	Output formatting (tables, trees, colors, timestamps)
`validators`	Input validation (emails, URLs, dates, pagination)
`template_engine`	Template loading and placeholder replacement
`project_detector`	Find existing Assistant Skills projects
`cache`	Response caching with TTL and LRU eviction
`error_handler`	Exception hierarchy and `@handle_errors` decorator

from assistant_skills_lib import (
    format_table, format_tree,
    validate_email, validate_url,
    Cache, handle_errors
)

@handle_errors
def fetch_data(resource_id):
    return api.get(f"/resources/{resource_id}")

📦 assistant-skills-lib on PyPI

Architecture

flowchart TD
    U["User Request"] --> CC["Claude Code"]
    CC --> AS["Assistant Skills<br/>Plugin"]

    AS --> SW["setup-wizard<br/>Environment Setup"]
    AS --> AB["assistant-builder<br/>Project Scaffolding"]
    AS --> SO["skills-optimizer<br/>Token Efficiency"]
    AS --> LP["landing-page<br/>README Branding"]
    AS --> LIB["library-publisher<br/>PyPI Publishing"]
    AS --> E2E["e2e-testing<br/>E2E Test Infrastructure"]

    SW --> VE["Shared Venv<br/>~/.assistant-skills-venv"]
    AB --> T["Templates"]
    AB --> SH["Shared Library"]

    SO --> AN["Analyzers"]
    LP --> TM["Template Engine"]
    LIB --> PY["PyPI Package<br/>Generation"]
    E2E --> DOC["Docker + pytest"]

Progressive Disclosure Model

Skills use 3 levels to minimize token usage:

Level	Target	Loaded When
L1: Metadata	~200 chars	Startup (all skills)
L2: SKILL.md	<500 lines	Skill triggered
L3: Nested docs	Variable	Explicitly accessed

Reference Projects

Templates derived from production implementations:

Project	Skills	Tests	Status
Jira-Assistant-Skills	14	560+	Production
Confluence-Assistant-Skills	14	1,039	Production
Splunk-Assistant-Skills	13	248+	Production

Development

Run Tests

pip install -r requirements.txt

# Run all skill tests (221 tests)
pytest skills/*/tests/ -v

# Run specific skill tests
pytest skills/assistant-builder/tests/ -v

Run E2E Tests

E2E tests validate the plugin by interacting with the actual Claude Code CLI:

# Run in Docker (requires ANTHROPIC_API_KEY)
./scripts/run-e2e-tests.sh

# Run locally (prefers OAuth, falls back to API key)
./scripts/run-e2e-tests.sh --local

Authentication:

Local runs: Prefers OAuth credentials (~/.claude.json). Run claude auth login first.
Docker runs: Requires ANTHROPIC_API_KEY environment variable.

Failed test responses are logged to test-results/e2e/responses_latest.log.

See tests/e2e/README.md for details.

Project Structure

Assistant-Skills/
├── .claude-plugin/
│   ├── plugin.json           # Plugin manifest
│   ├── marketplace.json      # Marketplace registry
│   ├── commands/             # Slash commands
│   │   ├── assistant-skills-setup.md
│   │   └── assistant-builder-setup.md
│   └── agents/               # Skill reviewer agents
├── skills/
│   ├── setup-wizard/         # Environment setup
│   ├── assistant-builder/    # Project scaffolding
│   ├── skills-optimizer/     # Token optimization
│   ├── landing-page/         # README branding
│   ├── library-publisher/    # PyPI publishing
│   └── e2e-testing/          # E2E test infrastructure
├── hooks/                    # Plugin hooks
│   └── hooks.json            # SessionStart health checks
├── docker/                   # Docker infrastructure
│   ├── runtime/              # Claude Code runtime image
│   └── e2e/                  # E2E test image
├── .github/workflows/        # GitHub Actions
│   └── docker-publish.yml    # Publish image on release
├── templates/                # Project templates
│   ├── 00-project-lifecycle/
│   ├── 01-project-scaffolding/
│   └── ...
├── tests/                    # E2E and integration tests
│   └── e2e/                  # End-to-end tests
├── pytest.ini                # Pytest configuration
├── conftest.py               # Shared test fixtures
├── requirements.txt          # Python dependencies
└── README.md

Docker Usage

Run Claude Code with Assistant Skills in a Docker container. No local Python or Node.js installation required.

Quick Start

# Pull and run (API key authentication)
export ANTHROPIC_API_KEY=sk-ant-...
./scripts/claude-as-docker.sh

# Or use OAuth (mount existing credentials)
./scripts/claude-as-docker.sh --oauth

Environment Variables

Variable	Description
`ANTHROPIC_API_KEY`	API key for Claude authentication
`CLAUDE_PLUGINS`	Comma-separated plugin repos (`owner/repo` or full URLs)
`CLAUDE_MARKETPLACES`	Comma-separated marketplace repos
`CLAUDE_REFRESH_PLUGINS`	Set to `false` to skip plugin updates (default: `true`)

Examples

# Run with a prompt
./scripts/claude-as-docker.sh -- -p "Help me refactor this code"

# Install additional plugins
CLAUDE_PLUGINS="myorg/my-plugin,other/plugin" ./scripts/claude-as-docker.sh

# Use multiple marketplaces
CLAUDE_MARKETPLACES="grandcamel/Assistant-Skills,company/internal-skills" \
  ./scripts/claude-as-docker.sh

# Skip plugin updates for faster startup
./scripts/claude-as-docker.sh --no-refresh

# Start a shell for debugging
./scripts/claude-as-docker.sh --shell

# Pull latest image before running
./scripts/claude-as-docker.sh --pull

Building Locally

# Build the image
docker build -t assistant-skills -f docker/runtime/Dockerfile .

# Run with custom image
./scripts/claude-as-docker.sh --image assistant-skills

Contributing

Contributions welcome!

# Clone the repository
git clone https://github.com/grandcamel/Assistant-Skills.git
cd Assistant-Skills

# Install dependencies and run tests
pip install -r requirements.txt
pytest skills/*/tests/ -v

License

MIT License — see LICENSE for details.

Stop writing boilerplate. Start building skills.
_{Built for Claude Code by developers who got tired of reinventing the wheel.}

Name		Name	Last commit message	Last commit date
Latest commit History 99 Commits
.claude-plugin		.claude-plugin
.claude		.claude
.github/workflows		.github/workflows
assets		assets
docker		docker
examples		examples
hooks		hooks
prompts		prompts
scripts		scripts
skills		skills
templates		templates
test_entrypoint		test_entrypoint
tests/e2e		tests/e2e
.dockerignore		.dockerignore
.env.template		.env.template
.gitignore		.gitignore
CLAUDE.md		CLAUDE.md
LICENSE		LICENSE
PHASE2_FEEDBACK.md		PHASE2_FEEDBACK.md
PHASE2_FEEDBACK_PART2.md		PHASE2_FEEDBACK_PART2.md
PHASE2_FEEDBACK_PART3.md		PHASE2_FEEDBACK_PART3.md
PROPOSAL_FEEDBACK.md		PROPOSAL_FEEDBACK.md
README.md		README.md
REFACTORING_FEEDBACK.md		REFACTORING_FEEDBACK.md
REFACTORING_INSTRUCTIONS.md		REFACTORING_INSTRUCTIONS.md
RESPONSE01.md		RESPONSE01.md
SANDBOX_TEST_RESULTS.md		SANDBOX_TEST_RESULTS.md
TEMPLATE-GUIDE.md		TEMPLATE-GUIDE.md
conftest.py		conftest.py
pytest.ini		pytest.ini
requirements-e2e.txt		requirements-e2e.txt
requirements.txt		requirements.txt

License

grandcamel/Assistant-Skills

Folders and files

Latest commit

History

Repository files navigation

Assistant Skills

10x

6

80+

0

The Difference

Starting from Scratch

With Assistant Skills

Time Saved

Quick Start

1. Install Plugin

2. Set Up Environment

Alternative: Docker

3. Create Your First Project

4. Start Building

Included Skills

setup-wizard

assistant-builder

skills-optimizer

landing-page

library-publisher

e2e-testing

Who Is This For?

Templates

Shared Library

Architecture

Progressive Disclosure Model

Reference Projects

Development

Run Tests

Run E2E Tests

Project Structure

Docker Usage

Quick Start

Environment Variables

Examples

Building Locally

Contributing

License

About

Topics

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases 3

Packages 0

Uh oh!

Contributors 2

Uh oh!

Languages

Packages