SkillBroker AutoGPT Plugin

Add expert knowledge to your AutoGPT agents with SkillBroker - the marketplace where AI agents pay for human expertise.

Installation

Option 1: pip install

pip install skillbroker-autogpt

Option 2: Clone into plugins directory

cd Auto-GPT/plugins
git clone https://github.com/skillbroker/skillbroker-autogpt.git

Configuration

Add to your .env file:

# Enable the plugin
ALLOWLISTED_PLUGINS=skillbroker-autogpt

# Optional: API key for authenticated requests
SKILLBROKER_API_KEY=your-api-key

# Optional: Override the API URL
SKILLBROKER_API_URL=https://api.skillbroker.io

Available Commands

Once installed, AutoGPT will have access to these commands:

skillbroker_search

Search the SkillBroker marketplace for expert skills.

Command: skillbroker_search
Args: {"query": "tax advice", "category": "Finance"}

skillbroker_invoke

Query a specific skill by ID.

Command: skillbroker_invoke
Args: {"skill_id": "freelancer-tax-advisor", "query": "Can I deduct my home office?"}

skillbroker_expert

Automatically find and query the best skill for your question.

Command: skillbroker_expert
Args: {"query": "What's the best way to structure a holding company?"}

Example Usage

When AutoGPT needs expert knowledge, it can use these commands:

Thinking: I need expert advice on tax deductions for freelancers.
I will use the skillbroker_expert command to get professional guidance.

Command: skillbroker_expert
Args: {"query": "What home office expenses can a freelancer deduct?"}

Result: [Expert: Freelancer Tax Advisor]

As a freelancer, you can deduct the following home office expenses:
1. Dedicated workspace - proportional rent/mortgage
2. Utilities - electricity, internet, heating
3. Office supplies and equipment
4. Business phone line
...

Using the Client Directly

You can also use the SkillBroker client in your own AutoGPT plugins:

from skillbroker_autogpt import SkillBrokerClient

client = SkillBrokerClient()

# Search for skills
results = client.search("machine learning")
for skill in results.skills:
    print(f"- {skill.name}: {skill.description}")

# Invoke a skill
response = client.invoke("skill-id", "Your question here")
print(response.response)

Plugin Structure

skillbroker_autogpt/
├── src/skillbroker_autogpt/
│   ├── __init__.py      # Package exports
│   ├── plugin.py        # AutoGPT plugin class
│   ├── client.py        # SkillBroker API client
│   └── models.py        # Data models
├── pyproject.toml
└── README.md

API Reference

SkillBrokerPlugin

The main plugin class that integrates with AutoGPT.

Commands:

• search_skills(query, category) - Search for skills
• invoke_skill(skill_id, query) - Invoke a specific skill
• ask_expert(query) - Auto-find and invoke the best skill

SkillBrokerClient

client = SkillBrokerClient(api_url=None, api_key=None, timeout=30)

# Methods
client.search(query, category, limit) -> SearchResult
client.invoke(skill_id, query) -> SkillResponse
client.get_skill(skill_id) -> Skill
client.get_recommendations(task_description, limit) -> List[Skill]

Support

• Documentation: https://skillbroker.io/docs
• API Reference: https://api.skillbroker.io/swagger
• Issues: https://github.com/skillbroker/skillbroker-autogpt/issues
• Email: contact@skillbroker.io

License

MIT License - see LICENSE file for details.