Configuration

Configure environment variables, API keys, and blockchain access for your prediction market agents

Environment file setup

The project uses environment variables for configuration. Create a .env file in the root directory of the repository.

1. Create .env file

   In the root of the repository, create a new file named .env:

   touch .env
   
2. Add required variables

   At minimum, add these two required variables:

   .env

   BET_FROM_PRIVATE_KEY=your_private_key_here
   OPENAI_API_KEY=your_openai_api_key_here
   

   Never commit your .env file to version control. It should already be in .gitignore.

3. Add optional variables

   Depending on which agents you want to run, you may need additional variables. The agent will inform you if any are missing when you try to run it.

Required variables

These variables are required for basic agent functionality:

BET_FROM_PRIVATE_KEY · string · required

Private key for the Ethereum wallet that will place bets on prediction markets.

Keep this key secure! It controls access to your funds on-chain. Use a dedicated wallet for testing with small amounts.

How to get it:

• Export from MetaMask or another wallet
• Generate a new wallet using web3 libraries
• Use a test wallet for development

OPENAI_API_KEY · string · required

API key for OpenAI services, used by agents that leverage GPT models for reasoning and predictions.

How to get it:

1. Sign up at platform.openai.com
2. Navigate to API keys section
3. Create a new secret key
4. Copy and save it securely

Optional API keys

These variables enable additional functionality:

Web search and research

SERPER_API_KEY · string

API key for Serper, which provides Google Search results for agent research.

Used by: Research agents, think thoroughly agents

Get it at: serper.dev

TAVILY_API_KEY · string

API key for Tavily, a search API optimized for AI agents and LLMs.

Used by: Research agents, web scraping agents

Get it at: tavily.com

Blockchain and data

GRAPH_API_KEY · string

API key for The Graph, used to query blockchain data from subgraphs.

Used by: Omen agents, market data fetching

Get it at: thegraph.com

Social media integration

FARCASTER_PRIVATE_KEY · string

Private key for Farcaster account integration.

Used by: Social media agents

TWITTER_ACCESS_TOKEN · string

Twitter OAuth access token for posting and reading tweets.

Used by: Social media agents

TWITTER_ACCESS_TOKEN_SECRET · string

Twitter OAuth access token secret.

Used by: Social media agents

TWITTER_BEARER_TOKEN · string

Twitter API v2 bearer token for authentication.

Used by: Social media agents

TWITTER_API_KEY · string

Twitter API consumer key.

Used by: Social media agents

TWITTER_API_KEY_SECRET · string

Twitter API consumer secret.

Used by: Social media agents

Get Twitter credentials at: developer.twitter.com

Monitoring and observability

LANGFUSE_SECRET_KEY · string

Secret key for Langfuse, an LLM engineering platform for tracing and monitoring.

Used by: Agent monitoring, debugging

LANGFUSE_PUBLIC_KEY · string

Public key for Langfuse.

Used by: Agent monitoring, debugging

LANGFUSE_HOST · string

Langfuse host URL (defaults to cloud instance if not specified).

Used by: Agent monitoring, debugging

Get Langfuse credentials at: langfuse.com

Vector database

PINECONE_API_KEY · string

API key for Pinecone vector database, used for semantic search and memory.

Used by: Long-term memory agents, embedding agents

Get it at: pinecone.io

Database

SQLALCHEMY_DB_URL · string

Database connection URL for SQLAlchemy (e.g., PostgreSQL, SQLite).

Format: postgresql://user:password@host:port/database

Used by: Data persistence, agent state management

Development mode

free_for_everyone · string · default: "0"

Set to 1 to open Streamlit apps in no-auth mode for development.

free_for_everyone=1

Only use this in development environments, not in production.

Complete .env.example

Here’s the complete .env.example file from the repository:

.env.example

OPENAI_API_KEY=
SERPER_API_KEY=
TAVILY_API_KEY=
BET_FROM_PRIVATE_KEY=
GRAPH_API_KEY=
LANGFUSE_SECRET_KEY=
LANGFUSE_PUBLIC_KEY=
LANGFUSE_HOST=
FARCASTER_PRIVATE_KEY=
PINECONE_API_KEY=
TWITTER_ACCESS_TOKEN=
TWITTER_ACCESS_TOKEN_SECRET=
TWITTER_BEARER_TOKEN=
TWITTER_API_KEY=
TWITTER_API_KEY_SECRET=
SQLALCHEMY_DB_URL=
# Opens streamlit in no-auth mode
free_for_everyone=1

Configuration by agent type

Different agents require different configurations:

Basic agents

Required for: coinflip, knownoutcome

BET_FROM_PRIVATE_KEY=your_private_key
OPENAI_API_KEY=your_openai_key

Research agents

Required for: think_thoroughly, prophet_* agents

BET_FROM_PRIVATE_KEY=your_private_key
OPENAI_API_KEY=your_openai_key
TAVILY_API_KEY=your_tavily_key
SERPER_API_KEY=your_serper_key

Social media agents

Required for: social_media

BET_FROM_PRIVATE_KEY=your_private_key
OPENAI_API_KEY=your_openai_key
TWITTER_ACCESS_TOKEN=your_token
TWITTER_ACCESS_TOKEN_SECRET=your_secret
TWITTER_BEARER_TOKEN=your_bearer
TWITTER_API_KEY=your_api_key
TWITTER_API_KEY_SECRET=your_api_secret
FARCASTER_PRIVATE_KEY=your_farcaster_key

Advanced agents

Required for: microchain_* agents with long-term memory

BET_FROM_PRIVATE_KEY=your_private_key
OPENAI_API_KEY=your_openai_key
PINECONE_API_KEY=your_pinecone_key
SQLALCHEMY_DB_URL=postgresql://user:pass@host/db
LANGFUSE_SECRET_KEY=your_langfuse_secret
LANGFUSE_PUBLIC_KEY=your_langfuse_public

Verifying configuration

When you run an agent, it will automatically check for required environment variables:

python prediction_market_agent/run_agent.py coinflip omen

If any required variables are missing, you’ll see an error message indicating which ones need to be set.

The agent will only check for variables it actually needs, so you don’t need to configure everything upfront.

Security best practices

Use test wallets

For development, use dedicated test wallets with small amounts of funds

Never commit secrets

Ensure .env is in .gitignore and never commit API keys or private keys

Rotate keys regularly

Regularly rotate API keys and private keys, especially after development

Use environment separation

Keep separate .env files for development, staging, and production

Next steps

Quickstart

Run your first agent with the coinflip example

Create custom agents

Build your own prediction market agent