AgentVerse CLI

Publish, discover, and manage AI skills, agents, workflows, souls and prompts from the AgentVerse marketplace. Use when working with the agentverse CLI to se...

Security Scan

The skill is internally consistent with its stated purpose (it simply documents and drives the agentverse CLI); the only notable risk is the documented install method which pipes remote GitHub-hosted scripts into a shell/PowerShell — a common but inherently risky pattern you should review before running.

Purpose & Capability

Name/description match the instructions and file metadata: the skill is an instruction-only wrapper for the agentverse CLI (search/publish/manage marketplace artifacts). Required binary 'agentverse' and the documented commands align with that purpose.

Instruction Scope

SKILL.md instructs the agent to run agentverse CLI commands and to set optional env vars (AGENTVERSE_URL, AGENTVERSE_TOKEN). It does not request unrelated files, system paths, or additional secrets beyond what the CLI would reasonably need.

Install Mechanism

The provided install instructions use curl from raw.githubusercontent.com piped to bash (and a PowerShell iex command). GitHub raw is a common release location and the skill.json also references GitHub releases, but piping remote scripts to a shell executes remote code and should be reviewed before running. This is expected for a CLI installer but is an elevated risk compared with instruction-only skills that require an already-installed binary.

Credentials

No required environment variables are declared. The docs mention AGENTVERSE_URL and AGENTVERSE_TOKEN as expected CLI flags/envs. Suggest caution if asked to provide unrelated tokens (e.g., GitHub tokens) to the tool; those are optional for rate-limit avoidance in self-update.

Persistence & Privilege

Skill is instruction-only, has always: false, and does not request persistent privileges or modify other skills or system-wide agent settings. Autonomous invocation is permitted by platform default but not combined with other red flags.

Assessment

This skill is essentially documentation for the agentverse CLI and is coherent with that purpose. The main thing to watch for: the install commands run a remote installer from raw.githubusercontent.com and execute it directly (curl | bash and PowerShell iex). That will run arbitrary code from that repository on your machine. Before installing, manually inspect the referenced install.sh / install.ps1 in the project repo (https://github.com/loonghao/agentverse) or prefer installing from signed GitHub release assets or your distro/package manager. Only provide tokens (GitHub, AGENTVERSE_TOKEN) if you understand why the CLI needs them and trust the upstream project. If you already have the 'agentverse' binary installed by a trusted channel, you can use this skill without running the remote installer.

AgentVerse (agentverse) is the CLI for the universal AI agent marketplace — publish, discover, and manage skills, agents, workflows, souls and prompts.

Installation

Linux / macOS:

curl -fsSL https://raw.githubusercontent.com/loonghao/agentverse/main/install.sh | bash

Windows (PowerShell):

irm https://raw.githubusercontent.com/loonghao/agentverse/main/install.ps1 | iex

What is AgentVerse?

Kind	Description
skill	Reusable capabilities (tools, functions)
agent	Autonomous AI agents with defined capabilities
workflow	Multi-step orchestration pipelines
soul	Persona and personality configurations
prompt	Optimized prompt templates

Quick Reference

Discovery

# Search across all artifact kinds
agentverse search --query "code review"
agentverse search --query "python" --kind skill

# List by kind / namespace
agentverse list --kind agent
agentverse list --kind skill --namespace myorg

# Get a specific artifact (latest version)
agentverse get --kind skill --namespace myorg --name my-skill

# Pin to a specific version
agentverse get --kind skill --namespace myorg --name my-skill --version 1.2.0

# Show version history
agentverse versions --kind skill --namespace python-tools --name linter

Publishing

# Publish a new artifact or new version
agentverse publish --file skill.toml

# Update metadata / content
agentverse update --kind skill --namespace myorg --name my-skill --file skill.toml

# Fork an artifact
agentverse fork --kind skill --namespace source-org --name original \
  --new-namespace myorg --new-name my-fork

# Deprecate (soft delete)
agentverse deprecate --kind skill --namespace myorg --name old-skill

Authentication

# Point to a custom AgentVerse server
export AGENTVERSE_URL=https://agentverse.example.com
agentverse login

# Register a new account
agentverse register --username alice --email [e-mail]

# Show current user
agentverse whoami

Social

# Rate an artifact (1–5 stars)
agentverse rate --kind skill --namespace myorg --name my-skill --stars 5

# Like / unlike
agentverse like   --kind skill --namespace myorg --name my-skill
agentverse unlike --kind skill --namespace myorg --name my-skill

# Post a comment
agentverse comment --kind skill --namespace myorg --name my-skill \
  --message "Great tool!"

# View social stats
agentverse stats --kind skill --namespace myorg --name my-skill

Agent Use (Programmatic)

# Record a learning insight
agentverse learn --kind skill --namespace myorg --name my-skill \
  --insight "Works well for Python 3.12"

# Submit benchmark results
agentverse benchmark --kind skill --namespace myorg --name my-skill \
  --score 0.95 --metric accuracy

Self-Update

# Check for newer version without installing
agentverse self-update --check

# Update to the latest release
agentverse self-update

# Use a GitHub token to avoid rate limits
agentverse self-update --token ghp_your_token

Global Flags

Flag	Env Var	Default	Description
--server	AGENTVERSE_URL	http://localhost:8080	Server URL
--token	AGENTVERSE_TOKEN	—	Bearer token for authenticated ops

Links

• Repository: https://github.com/loonghao/agentverse
• Docker Image: ghcr.io/loonghao/agentverse:latest