Overview

Dex Trove documentation

Dex Trove is a licence-clean library of agent artifacts: skills, agents, instructions, prompts, hooks, agentic workflows and plugins. It learns them from open repositories, admits only what may legally be redistributed, normalises each one into a standard folder with full provenance, and ships the result as a Python package (pip install dex-trove, CLI dex).

It is the content layer for Ananke Plexus. Ananke is the harness: registry, resolution, policy, activation and evidence. Dex Trove is the library you load into it. Dex also installs straight into Claude Code and GitHub Copilot, so you don’t need Ananke to use it.

flowchart LR
  U[Open repos<br/>awesome-copilot · anthropic · superpowers] --> L[dex learn<br/>licence · guards · normalise]
  F[First-party skills] --> L
  L --> T[(trove/<br/>970 items)]
  T --> P[PyPI: dex-trove]
  P --> A[Ananke registry]
  P --> C[Claude Code]
  P --> G[GitHub Copilot]
  T --> W[This site]

Start here

If you want to…	Read
Install the package and use your first skill	Getting started
Put items into Claude Code, Copilot or a folder	Installing items
Load the library into Ananke	Ananke integration
Understand why an item is (or isn’t) in the trove	Licence policy
Learn from a new repository	Adding a source
Look up a command or file format	CLI reference · Item format

Concepts

Why Dex Trove: the problem, and the principles that shape the solution.
Architecture: the learn pipeline, the trove, the package and the site.
Artifact kinds: skills, agents, instructions, prompts, hooks, workflows and plugins, and how each tool consumes them.
Licence policy: allow/review/deny, resolution order, attribution.
Provenance & versioning: pinned commits, content hashes, item SemVer.
Security model: untrusted content, guards, secret flags, scanning.

Guides

Reference

Design

Edit this page on GitHub ↗