APIs and tools designed for autonomous agents, not humans. Provably fair gambling, disposable identities, and the primitives agents need to operate independently.
$ pip install purpleflea
We don't build UIs. We build APIs, SDKs, and MCP servers that AI agents consume directly. No human in the loop required.
Every product is an API endpoint. No dashboards, no click-ops. Agents call our tools directly via REST, SDK, or MCP protocol.
We build the primitives agents need but don't have: fair randomness, disposable identities, verifiable computation. Infrastructure for autonomy.
Every operation produces a proof. Agents don't trust — they verify. Provably fair outcomes, signed responses, and auditable trails.
Two products. One mission. Give AI agents the infrastructure to operate independently in the real world.
Provably fair gambling for AI agents
A complete gambling API with cryptographic fairness proofs. Coin flips, dice rolls, custom odds, Kelly Criterion sizing, and instant crypto settlements. Every outcome is independently verifiable.
from purpleflea import AgentCasino casino = AgentCasino("sk_live_...") # Provably fair coin flip with Kelly sizing bet = casino.kelly_bet( bankroll=1.5, # ETH edge=0.02, odds=2.0 ) result = casino.flip( wager=bet.optimal_size, currency="ETH" ) # Verify fairness independently assert result.verify_proof() # True print(result.payout) # 0.061 ETH
Disposable identities for AI agents
On-demand email addresses and phone numbers that agents use to sign up for services, receive verification codes, and operate autonomously. Auto-extracts OTPs and cleans up after itself.
from purpleflea import BurnerIdentity burner = BurnerIdentity("sk_live_...") # Create disposable email + phone identity = burner.create( email=True, phone="US", ttl=3600 # auto-destruct in 1hr ) # Sign up for a service signup(email=identity.email) # Auto-extract verification code code = identity.wait_for_otp( timeout=30 ) print(code) # "847291" # Clean up identity.destroy()
Watch an AI agent use Purple Flea to place a provably fair bet and verify the outcome.
From API key to autonomous agent operations in minutes.
Sign up via our API. No dashboard, no forms. Your agent gets a key programmatically.
pip install purpleflea or connect via MCP server. Works with any agent framework.
Your agent calls our tools directly. Flip coins, create identities, verify outcomes.
Every result comes with cryptographic proof. Settlements happen on-chain, instantly.
Everything you need to integrate Purple Flea into your agent stack.
Open-source SDKs, examples, and MCP server implementations. Star us and contribute.
github.com/purpleflea →Machine-readable documentation for LLMs and AI agents. Discover all available tools and endpoints.
purpleflea.com/llms.txt →Model Context Protocol server for Claude, GPT, and any MCP-compatible agent. Install in one line.
View MCP setup →