Decision Hub CLI
An open-source skill registry and manager for 40+ agents with security scans and automated evals, geared to data science and ML.
- 114 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Registry skills remain third-party code and can run commands on your machine.
- GitHub login gives the registry organization data for private skills.
Install
- Works in
Manual install
curl -LsSf https://astral.sh/uv/install.sh | sh && PATH="$HOME/.local/bin:$PATH" uv tool install dhub-cliThe first Quickstart command from the README installs `uv` and the CLI. It is a script from the internet, so read it before running. If `uv` is already installed, `uv tool install dhub-cli` is enough.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Registry skills remain third-party code and can run commands on your machine.
- GitHub login gives the registry organization data for private skills.
This is third-party code. Review the repository files before installing.
What it does
Decision Hub is an open registry and package manager for AI agent skills from PyMC Labs. The `dhub` CLI searches skills in natural language with `dhub ask`, installs them into Claude Code, Cursor, Codex, Gemini CLI, Windsurf and dozens of other agents, and publishes your own from a local folder or GitHub. On publish a skill is scanned for command injection and credential exfiltration, gets a trust grade from A to F and can run eval cases in a sandbox where an LLM judge scores the output. Server, web UI and CLI are MIT-licensed and the registry can be self-hosted.
Who it is for. Data scientists, library authors and teams who want a vetted, private skill registry instead of copy-pasting prompts.
Good fit when
- An agent needs skills for PyMC, Stan or `scikit-learn`.
- A team wants to share versioned skills inside the organization.
- A library author ships a skill alongside a package.
- A third-party skill needs a security grade and eval results before install.
Not a fit when
- You need an MCP server registry, since Decision Hub stores only skills.
- You want an agent runtime rather than a capability store.
- A public registry is not acceptable and you cannot host your own.
Example request
Find a Bayesian statistics skill for PyMC with `dhub ask` and install it into Claude Code.Limitations
Publishing and private skills require a GitHub login with `dhub login`. The public `hub.decision.ai` registry is run by the project authors and third-party skills remain third-party code: the A–F grade lowers risk but does not replace reading. Eval scores depend on an LLM judge and do not guarantee correctness for your task. The platform targets data science, so the public registry holds the most skills for statistics and ML.
How to disable. Remove installed skills from the agent directory, then run `uv tool uninstall dhub-cli` or `pipx uninstall dhub-cli`.
FAQ
How do I install a skill into all agents at once?
Run `dhub install <org>/<skill> --agent all`. The skill is installed once and linked to every detected agent.
How do I publish my own skill?
`dhub init my-skill` creates a scaffold and `dhub publish ./my-skill` publishes it. A GitHub URL also works, and the registry then republishes on new commits.
Can a skill be private?
Yes, the `--private` flag restricts a skill to a GitHub organization, and access for other organizations is granted separately.
What is the trust grade?
A class from A to F that the registry assigns after scanning for command injection, credential exfiltration and dangerous patterns.
Can the registry be self-hosted?
Yes, server, interface and CLI are MIT-licensed. The repository includes `docker-compose-local.yml` and a `Makefile` for running it.
Discussion
No questions or reviews yet.
Related
CLI toolsA CLI and manager skill: one skills library distributed to 70+ agents, with management done from the agent conversation
A skill registry for teams with security scanning and versioning; the MCP server finds and installs skills on request.
A CLI skill manager for AI agents: install from repositories, translate between agent formats and deploy into their configs
An open catalogue of over three thousand skills, each backed by a real project, with industry collections and categories.
A CLI and Python SDK to search, install, create and evaluate reusable agent skills from a shared library
The `askit` command line toolkit for agent skills: lint, security scan, tests, forging new ones and export to any agent.
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Decision Hub (this card) | no | 114 | |
| Agent Skills CLI | no | 194 | installs and updates skills from various sources; Decision Hub adds its own registry, security grading and eval cases |
| Agent Skill Manager | no | 974 | manages the agent's local skills; here a central registry with publishing and natural-language search |
README in short
The registry stands out for checks on every publish, sandboxed eval cases and fully open code. It targets people installing skills for data science.