skillsmith 0.6.4

The premier agentic skills library for professional AI engineering

skillsmith Documentation

skillsmith helps you prepare a project so AI coding tools can understand your project better and work more reliably.

This guide is written for beginners. You can copy and run commands as shown.

1) Install

pip install skillsmith

Optional MCP support:

pip install "skillsmith[mcp]"

If skillsmith is not found, use module mode:

python -m skillsmith --help

2) 60-Second Start

Run this inside your project folder:

skillsmith init --guided

Then do:

skillsmith recommend
skillsmith add <skill-name>
skillsmith sync
skillsmith align
skillsmith audit --strict

3) What You Get

After setup, skillsmith creates and manages:

• .agent/ project context and memory files
• AI-tool instruction files (for supported tools)
• skill installation and tracking (skills.lock.json)
• workflow/evaluation/trust utilities

4) Command Reference (Simple)

Project Setup

• skillsmith init: Create the skillsmith workspace in your project.
• skillsmith sync: Re-scan your project and refresh generated files.
• skillsmith align: Re-render managed files from your saved profile.
• skillsmith doctor: Health check for setup and environment.
• skillsmith audit: Full quality/trust/drift audit.
• skillsmith report: Human-readable project status summary.
• skillsmith profile: View or change project profile settings.

Skills

• skillsmith list: Show available skills.
• skillsmith discover <query>: Search for skills by keyword.
• skillsmith recommend: Suggest skills based on your project.
• skillsmith add <skill-name-or-url>: Install one skill.
• skillsmith update: Update installed local skills.
• skillsmith lint: Validate skill metadata and structure.
• skillsmith rebuild: Rebuild local catalog from skill files.
• skillsmith assets status: Show optional runtime asset availability.
• skillsmith assets bootstrap: Download/copy runtime assets into local cache.

Workflow and Evaluation

• skillsmith compose "<goal>": Generate a step-by-step workflow for a goal.
• skillsmith eval: Run evaluation and save metrics artifact.
• skillsmith budget: Show context/token budget usage.
• skillsmith context-index build: Build searchable project context index.
• skillsmith context-index query "<query>": Search ranked project context.
• skillsmith context ...: Alias for context-index.

Registry and Trust (Advanced / Team Use)

• skillsmith registry: Manage team registry entries/lifecycle.
• skillsmith registry-service: Run/sync local registry service API.
• skillsmith trust-service: Run/sync local trust service API.

MCP Server and Context Snapshots

• skillsmith serve: Start MCP server for tool integrations.
• skillsmith snapshot: Save/restore .agent context snapshots.
• skillsmith watch: Monitor context drift and staleness.

Maintenance

• skillsmith update: Update installed skills.

5) Beginner-Friendly Workflows

A) First time in a project

skillsmith init --guided
skillsmith recommend
skillsmith add <skill-name>
skillsmith audit

B) Before starting a new task

skillsmith sync
skillsmith align
skillsmith compose "build <your-goal>"

C) Before merge/release

skillsmith eval
skillsmith audit --strict
skillsmith report

6) Most Useful Help Commands

Show global help:

skillsmith --help

Show help for one command:

skillsmith <command> --help

Examples:

skillsmith init --help
skillsmith audit --help
skillsmith registry-service --help

7) Library vs CLI

skillsmith is a Python package, but the supported public interface is the CLI.

• Stable: skillsmith <command> and python -m skillsmith <command>
• Not guaranteed stable: importing internal modules like skillsmith.commands.*

8) Development

Run from source:

PYTHONPATH=src python -m skillsmith --help

Build package artifacts:

uv run --group dev python -m build

9) Current Version

• Package version: 0.6.4

License

MIT. See LICENSE.