Skip to main content

The premier operating system for AI coding agents — standardize, compose, and scale skills-as-code.

Project description

Skillsmith Hero Banner

⚡ Skillsmith: The Agentic OS

The $100M Infrastructure Layer for AI-Assisted Engineering

PyPI version License: MIT Stars Discord

Standardize instructions. Evolve skills. Orchestrate swarms.

OverviewCompetitive MoatQuick StartPillarsCommand Reference


🚀 The Vision

Skillsmith isn't just a linter or a template generator. It is the Intelligence Backbone for professional engineering teams in the era of Agentic AI.

While LLMs are powerful, they are context-poor and structurally drift. Skillsmith solves this by turning static codebase rules into dynamic, Self-Evolving Skills-as-Code. It bridges the gap between raw AI inference and production-grade software delivery.


🛡️ The Competitive Moat

Skillsmith v1.0 introduces features that place it at the absolute frontier of AI engineering:

🌍 Ecosystem Discovery (CLI Hub)

Stop hunting for prompts. Use skillsmith search to query the sovereign Ghost-Content index of 889+ skills directly from your terminal. See global categories, filter by tech stack, and install them instantly.

🐝 Multi-Agent Swarm Orchestration

Stop running in single-agent loops. Skillsmith can decompose massive goals into Parallelized Task Graphs (skillsmith swarm plan), assigning specialized roles (Orchestrator, Researcher, Implementer, Reviewer) to coordinate complex missions across multiple agent threads via skillsmith team-exec.

🧬 Autonomous Evolution (XSkill)

Your project instructions should learn from your wins. Use skillsmith evolve to capture structural engineering patterns from your Git history or raw source code. Skillsmith translates developer experience into repeatable, benchmarked agent skills automatically.

🌲 Recursive Thinking Trees

When a strategy fails, Skillsmith doesn't just retry. It uses recursive AND/OR Thinking Trees (skillsmith tree) to resolve goals, exploring multiple implementation paths and strategy fallbacks before a single line of bad code is written.


🏁 High-Velocity Setup

1. Install

pip install skillsmith

2. Initialize (Sovereign Ghost-Sync)

# Sync 889+ expert skill packs natively (No Node/NPX required!)
skillsmith init --bundle python-pro
# or guided setup with structural profiling
skillsmith init --guided

3. Search & Install (Global Ecosystem) 🌍

Explore 889+ expert skills archived in our sovereign ghost-content branch.

# Search for skills (v1.0.8+)
skillsmith search --categories
skillsmith search "authentication" --limit 5

# Install directly from the sovereign index
skillsmith add --remote awesome next-js-expert

4. Evolve & Evaluate (Phase 2.2) 🌌🧪

Skillsmith isn't static. It autonomously captures patterns from your Git history and scores your entire skill library (including Ghost Skills) against industry standards.

# Capture new patterns from recent Git history
skillsmith evolve capture --limit 50

# Benchmark and score your expert library
skillsmith evolve evaluate

This generates a Skill Leaderboard and detailed compatibility report in .agent/evals/.

4. Align & Certify

# Keeps your Claude commands, Cursor rules, and IDE instructions 100% in sync.
skillsmith align
skillsmith doctor

4. Deploy a Swarm

# Decompose a complex mission into a parallel multi-agent plan.
skillsmith swarm plan "Refactor the entire auth module for OAuth2 parity"

🏗️ Architectural Pillars

Pillar Focus Capability
Pillar 1: Structural Integrity Zero Drift Unified instructions across Claude, Cursor, Windsurf, Zencoder, Copilot.
Pillar 2: Autonomous Evolution Self-Improvement Skills that auto-learn from history and code structures (XSkill/EvoSkill).
Pillar 3: Team Intelligence Swarm Logic Role-based execution with structured handoffs and verification loops.

🤖 Power Commands for Top-Tier Performance

Skillsmith injects a specialized command surface into your AI tools:

  • /swarm — Decompose massive goals into parallelized agent graphs with integrated Thinking Tree artifacts in .agent/TREE.md.
  • /team-exec — Execute missions with structured roles (O-R-I-R) and stateful MISSION.md updates.
  • /evolve — Capture structural engineering patterns from experience.
  • /tree — Strategic recursive reasoning for deep goal resolution; now supports --output for sharing strategic paths.
  • /ready — High-speed pre-release sanity check (Git, Profile, Index).

📈 Roadmap: The Path to v2.0

  • v0.7: Content Gravity & 33+ Workflows
  • v0.9: Autonomous Evolution (XSkill Phase 1)
  • v1.0: Competitive Evolution (Swarms & Thinking Trees)
  • v2.0: The Collective — Decentralized skill sharing & autonomous agent economies.

📄 License

MIT © ApexIQ

Built by the Google Deepmind team for those who engineer at the speed of thought.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

skillsmith-1.0.9.tar.gz (324.4 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

skillsmith-1.0.9-py3-none-any.whl (465.0 kB view details)

Uploaded Python 3

File details

Details for the file skillsmith-1.0.9.tar.gz.

File metadata

  • Download URL: skillsmith-1.0.9.tar.gz
  • Upload date:
  • Size: 324.4 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.11.13

File hashes

Hashes for skillsmith-1.0.9.tar.gz
Algorithm Hash digest
SHA256 d7535de8d4a514456a695030dbe40cd4374bf9de164b9ed6a4229ca4219184e8
MD5 3b44af2b081256edce60f74e36f4c5c0
BLAKE2b-256 791c7cf5a764bc03510866acb7a9ab999252d4c5c5d7e013d0fc98b81fd2ae14

See more details on using hashes here.

File details

Details for the file skillsmith-1.0.9-py3-none-any.whl.

File metadata

  • Download URL: skillsmith-1.0.9-py3-none-any.whl
  • Upload date:
  • Size: 465.0 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.11.13

File hashes

Hashes for skillsmith-1.0.9-py3-none-any.whl
Algorithm Hash digest
SHA256 f339bccc0e5da4861d499f82dea0d81230cf07a683ff4b107efe9cf6385b6189
MD5 318f8b590d0646e670785018439e4fd5
BLAKE2b-256 2cc8aa8dbfed61cb79fdbffb72dc4d4c46df06c6fa9c96ca56aa703cf67db1e2

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page