Skills Reference

Fetch-skills ships two categories of skills: package skills (pick exactly one) and protocol skills (pick one or more). They compose automatically — protocol skills defer to whichever package skill is present for all environment-level concerns such as install commands, lockfile management, and .env setup.

All Skills at a Glance

Skill	Type	What it provides
uv-package	Package	uv install/run/lock, pyproject.toml, .env template
poetry-package	Package	Poetry install/run/lock, pyproject.toml, .env template
python-venv-package	Package	venv, pip, requirements.txt, .env template
no-package	Package	Code-only; runtime import docs, no install commands
chat-protocol	Protocol	ChatMessage, session lifecycle, Agentverse manifest
stripe-payment-protocol	Protocol	Stripe Checkout RequestPayment/CommitPayment, idempotency
fet-payment-protocol	Protocol	On-chain FET, cosmpy ledger verify, mainnet + testnet
payment-protocol	Combined	Stripe + FET in one agent, unified on_commit dispatcher

---

Package Skills

Pick exactly one package skill per project. It controls all environment-level guidance your AI coding assistant uses: install commands, lockfile conventions, run commands, and .env / .gitignore templates.

uv-package

For projects using the uv package manager.

What your AI assistant learns:

• uv add / uv sync / uv run commands
• pyproject.toml shape and dependency groups
• uv.lock rules and .gitignore template
• .env.example layout for secrets

# Example commands your assistant will use
uv add uagents
uv sync
uv run agent.py

Recommended default

uv-package is the recommended choice for new Fetch.ai projects. uv is significantly faster than pip and handles virtual environments automatically.

---

poetry-package

For projects using Poetry.

What your AI assistant learns:

• poetry add / poetry install / poetry run commands
• pyproject.toml shape for Poetry projects
• poetry.lock rules
• .env.example layout for secrets

# Example commands your assistant will use
poetry add uagents
poetry install
poetry run python agent.py

---

python-venv-package

For classic venv + pip workflows.

What your AI assistant learns:

• python3.11 -m venv .venv setup steps
• pip install and pip freeze conventions
• requirements.txt structure
• .env.example layout for secrets

# Example commands your assistant will use
python3.11 -m venv .venv
source .venv/bin/activate   # or .venv\Scripts\activate on Windows
pip install uagents
pip freeze > requirements.txt

---

no-package

Code-only mode — no install commands of any kind.

What your AI assistant learns:

• Required runtime imports only
• No lockfile or install guidance
• Suitable for environments where dependencies are pre-installed

Use this when your CI environment, container, or platform already has all packages available and you only want the agent and protocol logic documented.

---

Protocol Skills

Pick one or more protocol skills. They compose cleanly with any package skill and with each other.

chat-protocol

Core protocol skill for building Fetch.ai uAgents that speak the official Agent Chat Protocol.

What your AI assistant learns:

• ChatMessage, ChatAcknowledgement, TextContent message types
• Session lifecycle management
• Agentverse manifest registration
• Handler patterns and message routing

# Pattern your assistant will follow
from uagents_core.contrib.protocols.chat import (
    ChatMessage,
    ChatAcknowledgement,
    TextContent,
)

@agent.on_message(ChatMessage)
async def handle_message(ctx: Context, sender: str, msg: ChatMessage):
    await ctx.send(sender, ChatAcknowledgement(timestamp=datetime.now(), acknowledged_msg_id=msg.msg_id))
    # handle msg.content here

Foundation skill

chat-protocol is the foundation for all payment skills. If you select stripe-payment-protocol or fet-payment-protocol, your agent must also speak Chat Protocol.

---

stripe-payment-protocol

Adds Stripe Checkout-backed payment support to a chat-capable uAgent.

What your AI assistant learns:

• RequestPayment and CommitPayment message flows
• Stripe Checkout session creation
• Payment verification and idempotency
• Environment setup for Stripe keys

Required environment variables:

STRIPE_SECRET_KEY=sk_live_...
STRIPE_WEBHOOK_SECRET=whsec_...

Payment flow:

User sends ChatMessage
  → Agent sends RequestPayment (Stripe Checkout URL)
  → User completes Stripe payment
  → Stripe webhook fires CommitPayment
  → Agent verifies and delivers service

Requires chat-protocol

stripe-payment-protocol composes with chat-protocol. Select both, or see payment-protocol for the combined Stripe + FET skill.

---

fet-payment-protocol

Adds direct on-chain FET payments to a uAgent alongside Chat Protocol.

What your AI assistant learns:

• funds with fet_direct payment method
• RequestPayment with on-chain metadata
• cosmpy ledger verification
• RejectPayment / CompletePayment handling
• Idempotency on transaction hash
• Supports Fetch.ai mainnet (afet) and stable-testnet (atestfet)

Required environment variables:

AGENT_WALLET_ADDRESS=fetch1...
FETCHAI_NETWORK=mainnet   # or stable-testnet

Payment flow:

User sends ChatMessage
  → Agent sends RequestPayment (on-chain FET address + amount)
  → User broadcasts FET transaction on-chain
  → User sends CommitPayment with transaction hash
  → Agent verifies via cosmpy ledger query
  → Agent completes or rejects based on verification

Requires chat-protocol

fet-payment-protocol composes with chat-protocol. Select both, or see payment-protocol for the combined Stripe + FET skill.

---

payment-protocol

Combined skill: both Stripe Checkout and direct FET payments in one agent.

What your AI assistant learns:

• Single RequestPayment.accepted_funds advertising both payment methods
• One on_commit handler dispatching by Funds.payment_method
• Full Stripe + FET environment setup
• Everything from both stripe-payment-protocol and fet-payment-protocol

Use this when: the same agent must accept card or crypto — users can choose their preferred payment method.

Required environment variables:

STRIPE_SECRET_KEY=sk_live_...
STRIPE_WEBHOOK_SECRET=whsec_...
AGENT_WALLET_ADDRESS=fetch1...
FETCHAI_NETWORK=mainnet

Payment flow:

User sends ChatMessage
  → Agent sends RequestPayment (both Stripe URL + FET address)
  → User picks their method and pays
  → CommitPayment arrives with payment_method field
  → on_commit dispatcher routes to Stripe or FET handler
  → Agent verifies and delivers service

---

Skill Composition Examples

Package skill	Protocol skill(s)	Result
uv-package	chat-protocol	Basic chat agent with uv environment
poetry-package	stripe-payment-protocol + chat-protocol	Stripe-gated agent for Poetry projects
python-venv-package	payment-protocol	Dual Stripe + FET agent with venv setup
no-package	chat-protocol + fet-payment-protocol	Code-only on-chain payment agent
uv-package	payment-protocol	Full dual-payment agent with uv (recommended for new projects)