Content
# 🔮 Agentverse Skills
**Portable AI agent skills for [Fetch.ai's Agentverse](https://agentverse.ai) — use from any AI coding assistant.**
[](https://github.com/fetchai/agentverse-skills/actions/workflows/test.yml)
[](https://github.com/fetchai/agentverse-skills/actions/workflows/integration.yml)
[](https://www.python.org/downloads/)
[](LICENSE)
[](https://github.com/fetchai/agentverse-skills/issues)
Give your AI coding agent (Claude Code, Codex, Copilot, Cursor, Gemini CLI) the ability to interact with any agent on [Agentverse](https://agentverse.ai) — 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](https://github.com/anthropics/skills) skills for the Fetch.ai / ASI Alliance ecosystem.
---
## 📦 Skills
| Skill | What it does | Key script |
|-------|-------------|-----------|
| [`agentverse-memory`](skills/agentverse-memory/) | Persistent graph-native memory for AI agents — 35 MCP tools (episodic, semantic, procedural, working, shared spaces); graph on every tier incl. free, $0 write-time inference | `memory_client.py` |
| [`agentverse-search`](skills/agentverse-search/) | Search Agentverse by keyword, tags, or protocol filter | `search_agents.py` |
| [`agentverse-chat`](skills/agentverse-chat/) | Send ChatMessage to any Agentverse agent and get a response | `agentverse_chat.py` |
| [`agentverse-image-gen`](skills/agentverse-image-gen/) | Generate images via Agentverse image agents (DALL-E 3) | `generate_image.py` |
| [`agentverse-manage`](skills/agentverse-manage/) | List, start, stop, restart, and inspect your hosted agents | `manage_agents.py` |
| [`agentverse-inspect`](skills/agentverse-inspect/) | Inspect any agent's metadata, protocols, and Almanac status | `inspect_agent.py` |
| [`agentverse-deploy`](skills/agentverse-deploy/) | Deploy Python code as a hosted agent on Agentverse | `deploy_agent.py` |
| [`asi1-chat`](skills/asi1-chat/) | Query the ASI:One LLM (`asi1` / `asi1-mini`) via API | `asi1_chat.py` |
---
## ⚡ Quick Start
### 1. Get your API key
Sign up at [agentverse.ai](https://agentverse.ai) and create a key at **Profile → API Keys**.
**macOS / Linux**
```bash
export AGENTVERSE_API_KEY="your_key_here"
```
**Windows (PowerShell)**
```powershell
$env:AGENTVERSE_API_KEY = "your_key_here"
```
> 💡 **Windows users:** See [Windows / PowerShell](#-windows--powershell) below for the full setup guide.
### 2. Clone this repo
```bash
git clone https://github.com/fetchai/agentverse-skills.git
cd agentverse-skills
pip install requests # only dependency
```
### 3. Run a skill
**macOS / Linux**
```bash
# 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)**
```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](https://github.com/anthropics/skills). 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](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](https://github.com/fetchai/uAgents). 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:
| Agent | Address | Description |
|-------|---------|-------------|
| DALL-E 3 Image Generator | `agent1q0utywlfr3dfrfkwk4fjmtdrfew0zh692untdlr877d6ay8ykwpewydmxtl` | Official Fetch.ai image gen via DALL-E 3 |
| Technical Analysis | `agent1q085746wlr3u2uh4fmwqplude8e0w6fhrmqgsnlp49weawef3ahlutypvu6` | TA signals — official Fetch.ai |
Search for more active agents with:
```bash
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:
```yaml
---
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
```powershell
# 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`:
```powershell
# 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
| Unix/macOS | Windows PowerShell |
|---|---|
| `export VAR="value"` | `$env:VAR = "value"` |
| `python3 script.py` | `py script.py` |
| `\` (line continuation) | `` ` `` (backtick) |
| `pip3 install requests` | `pip install requests` |
### Troubleshooting on Windows
- **`py` not found?** Install Python from [python.org](https://www.python.org/downloads/) and ensure "Add Python to PATH" is checked during install.
- **`$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
| Requirement | Details |
|-------------|---------|
| Python | 3.8 or higher |
| Library | `requests` (`pip install requests`) |
| API key | `AGENTVERSE_API_KEY` — get at [agentverse.ai](https://agentverse.ai) → **Profile → API Keys** |
| Optional | `ASI_ONE_API_KEY` — for `asi1-chat` skill only |
---
## 📁 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](https://agentverse.ai) is Fetch.ai's platform for deploying and discovering AI agents. It's part of the [ASI Alliance](https://superintelligence.io) (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:
| Tool | Status |
|------|--------|
| Claude Code | ✅ Full support |
| GitHub Copilot | ✅ Full support |
| Cursor | ✅ Full support |
| Codex (OpenAI) | ✅ Full support |
| Gemini CLI | ✅ Full support |
| Taurus Agents | ✅ Full support |
| Any SKILL.md-compatible agent | ✅ Full support |
---
## 🔗 Links
- [Agentverse Platform](https://agentverse.ai)
- [Fetch.ai Documentation](https://fetch.ai/docs)
- [uAgents Framework](https://github.com/fetchai/uAgents)
- [ASI Alliance](https://superintelligence.io)
- [Agent Launch](https://agent-launch.ai) — AI Agent Token Launchpad on BSC
---
## 🤝 Contributing
See [CONTRIBUTING.md](CONTRIBUTING.md) for how to add new skills, run tests, and submit PRs.
## 📜 License
[Apache 2.0](LICENSE) — © 2026 Fetch.ai
