Skip to main content

The AI Skill Manager for Data Science Agents

Project description

dhub-cli: The AI Skill Manager

Decision Hub is a registry for publishing, discovering, and installing Skills — modular packages of code and prompts that AI coding agents (Claude, Cursor, Codex, Gemini, OpenCode) can use.

Browse the registry at hub.decision.ai or use the CLI below.

Why Decision Hub?

Agents that extend themselves. Install Decision Hub as a skill into any supported agent, and the agent can discover new skills in natural language — then install and use them mid-conversation without human intervention.

Publish from anywhere. Point dhub publish at a local directory or a GitHub repo URL and every valid SKILL.md inside is discovered, versioned, and published.

Private skills for your team. Skills can be scoped to your GitHub organization so proprietary tooling stays internal.

Install once, use everywhere. A single dhub install symlinks a skill into every agent's skill directory — Claude, Cursor, Codex, Gemini, OpenCode. No duplication, no per-agent setup.

Security gauntlet. Every publish is scanned for dangerous patterns. Skills get a trust grade (A/B/C/F) before they reach the registry.

Automated evals in sandboxes. Skills ship with eval cases that run on publish in isolated sandboxes, scored by an LLM judge.

Executable skills with SKILL.md. Builds on the Agent Skills spec with runtime and evals blocks — skills are runnable programs, not just static prompts.

Installation

# Via uv (recommended)
uv tool install dhub-cli

# Via pipx
pipx install dhub-cli

Quick Start

# 1. Authenticate via GitHub
dhub login

# 2. Search for skills using natural language
dhub ask "analyze A/B test results"

# 3. Install a skill for all your agents
dhub install pymc-labs/causalpy

# 4. Scaffold a new skill
dhub init my-new-skill

# 5. Publish it under your namespace
# (Run this inside the skill directory)
dhub publish .

Supported Agents

Skills are installed as symlinks into each agent's skill directory, making them immediately available:

  • Claude: ~/.claude/skills
  • Cursor: ~/.cursor/skills
  • Codex: ~/.codex/skills
  • Gemini: ~/.gemini/skills
  • OpenCode: ~/.config/opencode/skills

Documentation

For full documentation on creating skills, the SKILL.md format, and running your own registry server, see the main repository.

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

dhub_cli-0.10.1.tar.gz (52.7 kB view details)

Uploaded Source

Built Distribution

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

dhub_cli-0.10.1-py3-none-any.whl (38.1 kB view details)

Uploaded Python 3

File details

Details for the file dhub_cli-0.10.1.tar.gz.

File metadata

  • Download URL: dhub_cli-0.10.1.tar.gz
  • Upload date:
  • Size: 52.7 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: uv/0.6.12

File hashes

Hashes for dhub_cli-0.10.1.tar.gz
Algorithm Hash digest
SHA256 6561c582fff1d10697975ac7f0d8f4cdbd205079263bd6f9a9a2c36e3398f60b
MD5 c1bb1331940fcc9136cd2b5aae583fcd
BLAKE2b-256 9f9d36ce0709042f24b2003ffbcf1ade58a910ec16c93b6267ac27ddc4667c2d

See more details on using hashes here.

File details

Details for the file dhub_cli-0.10.1-py3-none-any.whl.

File metadata

  • Download URL: dhub_cli-0.10.1-py3-none-any.whl
  • Upload date:
  • Size: 38.1 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: uv/0.6.12

File hashes

Hashes for dhub_cli-0.10.1-py3-none-any.whl
Algorithm Hash digest
SHA256 766296af09fed24384082c3c255ba654e94d68f92aed37eb52c1078e7cf6cbf6
MD5 e9834a2d62829799eadc12a19af8819f
BLAKE2b-256 c2dae35d50868f8041536d30bce54f66d2a09b3561aaaadd69e5c5d5fac3e85b

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