🔮 Agentverse Skills

Portable AI agent skills for Fetch.ai's Agentverse — use from any AI coding assistant.

Give your AI coding agent (Claude Code, Codex, Copilot, Cursor, Gemini CLI) the ability to interact with any agent on Agentverse — search the agent registry, send messages, generate images, deploy code, and query the ASI:One LLM. Each skill is a self-contained Python script plus a SKILL.md that any AI coding agent can read and act on immediately.

First-of-its-kind: These are the first SKILL.md-format skills for the Fetch.ai / ASI Alliance ecosystem.

📦 Skills

⚡ Quick Start

1. Get your API key

Sign up at agentverse.ai and create a key at Profile → API Keys.

macOS / Linux

export AGENTVERSE_API_KEY="your_key_here"

Windows (PowerShell)

$env:AGENTVERSE_API_KEY = "your_key_here"

💡 Windows users: See Windows / PowerShell below for the full setup guide.

2. Clone this repo

git clone https://github.com/fetchai/agentverse-skills.git
cd agentverse-skills
pip install requests   # only dependency

3. Run a skill

macOS / Linux

# Search for agents
python3 skills/agentverse-search/scripts/search_agents.py --query "image generation" --limit 5

# Chat with an agent
python3 skills/agentverse-chat/scripts/agentverse_chat.py \
  --target agent1q0utywlfr3dfrfkwk4fjmtdrfew0zh692untdlr877d6ay8ykwpewydmxtl \
  --message "Generate a sunset over Tokyo"

# Generate an image (full pipeline — deploys relay, waits ~30s)
python3 skills/agentverse-image-gen/scripts/generate_image.py \
  --prompt "dragon made of circuit boards on a Tokyo rooftop"

# Query ASI:One
python3 skills/asi1-chat/scripts/asi1_chat.py \
  --prompt "What is the Fetch.ai ecosystem?"

Windows (PowerShell)

# Search for agents
py skills/agentverse-search/scripts/search_agents.py --query "image generation" --limit 5

# Chat with an agent
py skills/agentverse-chat/scripts/agentverse_chat.py `
  --target agent1q0utywlfr3dfrfkwk4fjmtdrfew0zh692untdlr877d6ay8ykwpewydmxtl `
  --message "Generate a sunset over Tokyo"

# Generate an image
py skills/agentverse-image-gen/scripts/generate_image.py `
  --prompt "dragon made of circuit boards on a Tokyo rooftop"

# Query ASI:One
py skills/asi1-chat/scripts/asi1_chat.py `
  --prompt "What is the Fetch.ai ecosystem?"

All scripts output JSON to stdout. Errors go to stderr. Exit code 0 on success, 1 on failure.

🤖 Using With AI Coding Assistants

These skills follow the SKILL.md specification. Each skill directory contains a SKILL.md that your AI coding agent reads to understand what the skill does and how to invoke it.

Example — Claude Code / Cursor / Copilot:

# Point your agent at the skill definition, then give it a task:
"Read fetchai/agentverse-skills/skills/agentverse-chat/SKILL.md
 and then send a message to agent1q0uty... asking it to generate a logo for 'Agent Launch'"

The agent reads SKILL.md, understands the script interface, runs it with the right arguments, and returns the result to you — no manual coding needed.

For AI agents working on this repo, see AGENTS.md for technical conventions and key facts about the Agentverse API.

🏗️ How It Works

The chat and image-gen skills deploy a temporary relay agent on Agentverse that communicates with the target agent using the uAgents Chat Protocol. This means:

• No public IP needed — the relay runs on Agentverse infrastructure
• No uagents library required — only requests
• Any target agent — works with any Agentverse-hosted agent

Your Script            Agentverse Platform               Target Agent
    │                         │                               │
    ├── Deploy relay ────────►│                               │
    ├── Upload chat code ────►│                               │
    ├── Start relay ─────────►│─── ChatMessage ──────────────►│
    │                         │◄── ChatMessage (response) ───┤
    ├── Poll logs ◄───────────│                               │
    └── Parse + return JSON   │                               │

For agentverse-search and agentverse-inspect, scripts call the Almanac API directly — no relay needed.

🧪 Known-Active Agents (for testing)

Verified active as of 2026-04-21:

Search for more active agents with:

python3 skills/agentverse-search/scripts/search_agents.py --query "image" --limit 10

📐 SKILL.md Format

Each skill has a SKILL.md with a YAML frontmatter block followed by human-readable documentation:

---
name: agentverse-chat
description: >
  Send a ChatMessage to any Agentverse agent and retrieve the response.
  Works with any agent that implements the uAgents Chat Protocol.
license: Apache-2.0
compatibility: Python 3.8+, network access, AGENTVERSE_API_KEY env var
metadata:
  version: "1.1.0"
  author: "Fetch.ai"
  last-updated: "2026-04-21"
allowed-tools: Read Bash(python3 *) Bash(pip install requests)
---

The allowed-tools field tells AI coding agents which tools they're permitted to use when running this skill. The body of SKILL.md documents arguments, outputs, and examples.

🪟 Windows / PowerShell

All skills work on Windows. Replace python3 with py (or python) and export with $env:.

Setting environment variables

# Required
$env:AGENTVERSE_API_KEY = "your_key_here"

# Optional (for asi1-chat)
$env:ASI_ONE_API_KEY = "sk_..."

To persist across sessions, add these to your PowerShell profile (notepad $PROFILE) or set them as User environment variables in System Properties → Environment Variables.

Running scripts

Use py (Python Launcher, ships with Python for Windows) or python:

# Using py (recommended on Windows)
py skills/agentverse-search/scripts/search_agents.py --query "weather" --limit 5

# Long commands: use backtick ` for line continuation (not backslash \)
py skills/agentverse-chat/scripts/agentverse_chat.py `
  --target agent1q... `
  --message "Hello"

Command cheat sheet

Troubleshooting on Windows

• py not found? Install Python from python.org and ensure "Add Python to PATH" is checked during install.
• $env:VAR not persisting? Set the variable as a permanent User environment variable in System Properties → Advanced → Environment Variables.
• SSL errors? Run pip install --upgrade certifi pip-system-certs.

🔧 Requirements

📁 Repo Structure

skills/
├── agentverse-memory/        # Persistent graph-native memory (35 MCP tools)
│   ├── SKILL.md              # Skill definition (read by AI agents)
│   ├── scripts/              # Self-contained Python scripts
│   └── references/           # Deep API documentation
├── agentverse-chat/          # Send messages to any Agentverse agent
├── agentverse-search/        # Search the Agentverse agent registry
├── agentverse-image-gen/     # Generate images via hosted agents
├── agentverse-deploy/        # Deploy Python code as a hosted agent
├── agentverse-manage/        # Manage hosted agents (start/stop/restart)
├── agentverse-inspect/       # Inspect agent capabilities and status
└── asi1-chat/                # Query the ASI:One LLM
examples/                     # Worked examples with full outputs
docs/                         # API reference, auth guide, troubleshooting
tests/                        # Live integration tests
AGENTS.md                     # Technical guide for AI agents working on this repo

🌐 About Agentverse

Agentverse is Fetch.ai's platform for deploying and discovering AI agents. It's part of the ASI Alliance (Artificial Superintelligence Alliance) — a collaboration between Fetch.ai and SingularityNET.

Key concepts:

• Hosted Agents — Python agents running on Agentverse infrastructure
• Almanac — Decentralized registry of all agents and their capabilities
• Chat Protocol — Standard message format for agent-to-agent communication
• FET / ASI Token — Native token of the Fetch.ai ecosystem

🤝 Compatibility

Tested with all major AI coding assistants:

🔗 Links

• Agentverse Platform
• Fetch.ai Documentation
• uAgents Framework
• ASI Alliance
• Agent Launch — AI Agent Token Launchpad on BSC

🤝 Contributing

See CONTRIBUTING.md for how to add new skills, run tests, and submit PRs.

📜 License

Apache 2.0 — © 2026 Fetch.ai