Skip to content

Releases: lhassa8/skillforge

v1.0.0: Production Release

22 Jan 19:57

Choose a tag to compare

v1.0.0 - Production Release

SkillForge is now production-ready! This release completes the roadmap from v0.9.0 to v1.0.0.

Installation

pip install ai-skillforge

Or with AI features:

pip install ai-skillforge[ai]

PyPI: https://pypi.org/project/ai-skillforge/1.0.0/

Added

  • Enterprise Configuration - Full enterprise deployment support

    • SSO authentication support (SAML, OIDC)
    • Cloud storage backends (S3, GCS, Azure Blob)
    • Proxy configuration with SSL verification options
    • Telemetry and logging configuration
    • Environment variable overrides with SKILLFORGE_ prefix
    • User and project-level configuration files
  • Configuration CLI Commands

    • skillforge config show - Display current configuration
    • skillforge config set <key> <value> - Set configuration values
    • skillforge config path - Show configuration file locations
    • skillforge config init - Create configuration files (user/project)
  • Migration Tools - Upgrade skills from older formats

    • Auto-detect skill format versions (v0.1, v0.9, v1.0)
    • skillforge migrate check ./skills - List skills needing migration
    • skillforge migrate run ./skills/my-skill - Migrate a skill to v1.0
    • skillforge migrate preview ./skills/my-skill - Preview changes
    • Automatic backup creation before migration
    • Batch directory migration support
  • Stable Public API - skillforge.api module

    • Comprehensive exports of all public classes and functions
    • Semantic versioning guarantees for all exported APIs
    • deprecated() decorator for managing API deprecations
  • Enhanced Commands

    • skillforge doctor - Check installation and environment
    • skillforge info - Show detailed SkillForge information

Changed

  • Updated default model to claude-sonnet-4-20250514
  • All public APIs are now stable with deprecation warnings before removal

Full Changelog

See CHANGELOG.md for complete details.

v0.8.0 - Skill Composition

22 Jan 14:17

Choose a tag to compare

Skill Composition

Combine multiple skills into composite skills by including other skills' content.

New Features

  • skillforge compose - Compose a skill by resolving includes into a single file
  • includes field in SKILL.md - Reference other skills to include their content
  • Circular dependency detection - Prevents infinite loops in skill composition
  • Auto-composition - bundle and preview commands handle composite skills automatically

Example

---
name: full-stack-reviewer
description: Comprehensive code review
includes:
  - ./components/frontend-review
  - ./components/backend-review
---

# Full Stack Reviewer

This skill combines frontend and backend review capabilities.

Install

pip install --upgrade ai-skillforge

v0.7.0 - Skill Registry

22 Jan 13:50

Choose a tag to compare

What's New

Skill Registry - Discover and download skills from GitHub-based registries.

New Commands

# Registry management
skillforge registry add https://github.com/user/skill-registry
skillforge registry list
skillforge registry remove my-registry
skillforge registry update

# Discover and download
skillforge search "code review"
skillforge pull code-reviewer

How Registries Work

A registry is a GitHub repo containing an index.json file:

{
  "name": "my-registry",
  "description": "A collection of useful skills",
  "skills": [
    {
      "name": "code-reviewer",
      "description": "Review code for best practices",
      "version": "1.0.0",
      "repo": "https://github.com/user/code-reviewer-skill",
      "tags": ["code", "review"]
    }
  ]
}

Programmatic API

from skillforge import add_registry, search_skills, pull_skill

# Add a registry
add_registry("https://github.com/skillforge/community-skills")

# Search for skills
results = search_skills("code review")
for skill in results:
    print(f"{skill.name}: {skill.description}")

# Download a skill
skill_dir = pull_skill("code-reviewer", output_dir=Path("./skills"))

Full Changelog: v0.6.0...v0.7.0

v0.6.0 - AI-Powered Skill Analysis

22 Jan 13:29

Choose a tag to compare

What's New

AI-Powered Skill Analysis - Get quality scores and improvement suggestions for your skills using AI.

New Features

  • skillforge analyze ./skills/my-skill - Analyze a skill for quality
  • skillforge analyze ./skills/my-skill --json - Output as JSON for CI/tooling
  • Quality scores (0-100): clarity, completeness, examples, actionability
  • Detailed feedback: strengths, suggestions, issues

Programmatic API

from skillforge import analyze_skill, AnalysisResult

result = analyze_skill(Path("./skills/my-skill"))
if result.success:
    print(f"Overall score: {result.overall_score}/100")
    print(f"Suggestions: {result.suggestions}")

Example Output

Skill Analysis: code-reviewer

Overall Score: 78/100

┏━━━━━━━━━━━━━━━━┳━━━━━━━━┓
┃ Dimension      ┃ Score  ┃
┡━━━━━━━━━━━━━━━━╇━━━━━━━━┩
│ Clarity        │ 85/100 │
│ Completeness   │ 70/100 │
│ Examples       │ 75/100 │
│ Actionability  │ 82/100 │
└────────────────┴────────┘

Strengths:
  • Clear step-by-step review process
  • Good coverage of common code issues

Suggestions:
  • Add examples for different languages
  • Include performance review guidelines

Full Changelog: v0.5.0...v0.6.0

v0.5.0 - Claude Code Integration

22 Jan 02:10

Choose a tag to compare

What's New

Claude Code Integration - Install skills directly to Claude Code

New Commands

# Install to user directory (~/.claude/skills/)
skillforge install ./skills/my-skill

# Install to project directory (./.claude/skills/)
skillforge install ./skills/my-skill --project

# Uninstall
skillforge uninstall my-skill

# Sync all skills from a directory
skillforge sync ./skills

# List installed skills
skillforge installed

How It Works

Claude Code automatically discovers skills from:

  • ~/.claude/skills/ - User-level (available in all projects)
  • ./.claude/skills/ - Project-level (project-specific)

SkillForge copies your skills to these directories, making them immediately available in Claude Code.

Programmatic API

from skillforge import install_skill, list_installed_skills, sync_skills

# Install a skill
result = install_skill('./skills/my-skill', scope='user')
print(f'Installed: {result.skill_name}')

# List installed skills
for skill in list_installed_skills():
    print(f'{skill.name} ({skill.scope}): {skill.description}')

# Sync all skills from a directory
installed, errors = sync_skills('./skills', scope='user')

Install

pip install ai-skillforge==0.5.0

Full Changelog: v0.4.0...v0.5.0

v0.4.0 - Skill Templates

22 Jan 01:56

Choose a tag to compare

What's New

Skill Templates - Start skills from built-in templates for common use cases

New Commands

# List available templates
skillforge templates

# Preview a template
skillforge templates show code-review

# Create skill from template
skillforge new my-reviewer --template code-review

Built-in Templates

Template Description
code-review Review code for best practices, bugs, security
git-commit Write conventional commit messages
git-pr Create comprehensive PR descriptions
api-docs Generate API documentation
debugging Systematic debugging assistance
sql-helper Write and optimize SQL queries
test-writer Generate unit tests
explainer Explain complex code or concepts

Programmatic API

from skillforge import get_template, list_templates, SkillTemplate

# Get a specific template
template = get_template('code-review')
print(template.content)

# List all templates
for tmpl in list_templates():
    print(f'{tmpl.name}: {tmpl.description}')

Install

pip install ai-skillforge==0.4.0

Full Changelog: v0.3.0...v0.4.0

v0.3.0 - Skill Testing Framework

21 Jan 22:55

Choose a tag to compare

Skill Testing Framework

Test your skills before deployment with mock or live modes.

New Command: skillforge test

# Basic mock mode testing
skillforge test ./skills/my-skill

# Live mode with real API calls
skillforge test ./skills/my-skill --mode live

# Filter by tags
skillforge test ./skills/my-skill --tags smoke,critical

# Output formats for CI
skillforge test ./skills/my-skill --format json -o results.json
skillforge test ./skills/my-skill --format junit -o results.xml

# Cost estimation
skillforge test ./skills/my-skill --mode live --estimate-cost

Test Definition Format (tests.yml)

version: "1.0"
tests:
  - name: "basic_test"
    input: "User input to test"
    assertions:
      - type: contains
        value: "expected text"
      - type: regex
        pattern: "(feat|fix):"
    trigger:
      should_trigger: true
    mock:
      response: "Mocked response"
    tags: ["smoke"]

Assertion Types

  • contains / not_contains - Text presence
  • regex - Pattern matching
  • starts_with / ends_with - Position matching
  • length - Response length bounds (min/max)
  • json_valid - Valid JSON check
  • json_path - JSONPath value assertion
  • equals - Exact match

Programmatic API

from skillforge import (
    load_test_suite,
    run_test_suite,
    TestResult,
    TestSuiteResult,
)

skill, suite = load_test_suite(Path("./skills/my-skill"))
result = run_test_suite(skill, suite, mode="mock")

print(f"Passed: {result.passed_tests}/{result.total_tests}")

Installation

pip install --upgrade ai-skillforge

Full Changelog

https://github.com/lhassa8/skillforge/blob/main/CHANGELOG.md

v0.2.2 - Security Fix

17 Jan 15:38

Choose a tag to compare

Security Fix

This release excludes sensitive files from package builds.

Fixed

  • Excluded .claude/ directory from package builds
  • Added sdist exclusion rules in pyproject.toml for sensitive files (.claude/, api_key.txt, skills/, *.zip)

Upgrade

pip install --upgrade ai-skillforge

v0.2.1 - Security & CI Improvements

14 Jan 21:52

Choose a tag to compare

What's Changed

Added

  • GitHub Actions CI workflow (Python 3.10-3.13 on Ubuntu and macOS)
  • Security section in README

Fixed

  • Path traversal protection in zip extraction
  • Symlinks excluded from bundles for security
  • API key validation before provider calls
  • Ruff and mypy linting errors
  • Programmatic usage docs now show correct types (Path objects, result classes)

Changed

  • Package renamed from skillforge to ai-skillforge on PyPI
  • Use importlib.util.find_spec for package detection

Install

pip install ai-skillforge
pip install ai-skillforge[ai]  # With AI generation

Full Changelog: v0.2.0...v0.2.1

v0.2.0 - AI-Powered Skill Generation

14 Jan 01:17

Choose a tag to compare

Anthropic Agent Skills Creator

SkillForge is now the developer toolkit for Anthropic Agent Skills — custom instructions that extend Claude's capabilities.

Installation

pip install ai-skillforge

# With AI-powered generation (recommended)
pip install ai-skillforge[ai]

Note: Package renamed from skillforge to ai-skillforge on PyPI.

Highlights

  • AI-Powered Generation — Generate complete skills from natural language descriptions
  • AI-Powered Improvement — Enhance existing skills with more examples and better instructions
  • Multi-Provider Support — Works with Anthropic, OpenAI, and Ollama
  • Security Hardened — Path traversal protection, symlink handling, API key validation

New Commands

Command Description
skillforge generate Generate a skill using AI
skillforge improve Improve an existing skill with AI
skillforge providers Show available AI providers
skillforge new Create a skill scaffold
skillforge validate Validate against Anthropic requirements
skillforge bundle Package for upload to claude.ai
skillforge show Display skill details
skillforge preview Preview how Claude sees the skill

Quick Start

# Set your API key
export ANTHROPIC_API_KEY=your-key

# Generate a complete skill from a description
skillforge generate "Help users write clear git commit messages"

# Validate and bundle
skillforge validate ./skills/my-skill
skillforge bundle ./skills/my-skill

Security Improvements

  • Path traversal protection in zip extraction
  • Symlinks excluded from bundles
  • API key validation before provider calls
  • Bundle size recommendations (10MB max)

Full Documentation

See the README for complete documentation including:

  • Command reference with all options
  • SKILL.md format specification
  • Programmatic usage examples
  • Troubleshooting guide