Prove Your AI AgentActually Works
Cryptographic attestations for AI agents. Every action your agent takes is signed, timestamped, and independently verifiable - no trust required.
How It Works
Sign any action with Ed25519, get a public verification URL, let anyone verify independently
Sign Any Action
Create cryptographically signed attestations for any action your AI agent takes
Features
from treeship_sdk import Treeship
ts = Treeship(api_key="...")
result = ts.attest(
action="Approved loan #12345",
agent="loan-agent"
)"Your agent attests it approved a loan - the signature proves the action happened at that exact time"
Public Verification Pages
Every attestation gets a shareable URL anyone can visit to verify authenticity
Features
# Share with customers
print(result.verify_url)
# https://treeship.dev/verify/
# loan-agent/abc-123...
# Anyone can verify
treeship verify abc-123..."Your customer clicks the link and sees proof the action was signed by your agent's key"
Independent Verification
Don't trust Treeship - verify attestations yourself using just the public key
Features
# Get the public key
curl api.treeship.dev/v1/pubkey
# Verify locally with OpenSSL
openssl pkeyutl -verify \
-pubin -inkey pubkey.pem \
-sigfile sig.bin -in payload.txt"Security teams can verify attestations without trusting Treeship's servers at all"
Treeship Platform
Enterprise-grade verification infrastructure for AI agents. Prove capabilities with cryptographic evidence.
API-First
RESTful APIs for seamless integration
Enterprise Security
SOC 2 compliant with end-to-end encryption
Real-time Attestations
Generate verifiable attestations instantly
Monitor Your Treeships
Analytics dashboard for agent performance tracking
Simple Integration
Works with any Python codebase. Add verification to existing functions without changing your architecture.
# agent.py - Real verification examples
from treeship import attest_memory, attest_reasoning
@attest_memory # Proves memory updates
def remember_user_context(user_id, context):
self.memory.update(user_id, context)
@attest_reasoning # Proves decision logic
def make_recommendation(user_data):
reasoning = analyze_preferences(user_data)
return generate_recommendation(reasoning)
# Customers verify: memory accuracy, reasoning steps, performanceHow It Works
From memory updates to cryptographic proofs in three simple steps
Agent Executes & Learns
Your AI agent processes requests, updates memory, and makes decisions. Every operation gets automatically tracked.
Automatic Attestation
Treeship automatically captures agent operations and generates verifiable attestations. Real-time verification without complexity.
Customer Verification
Your customers independently verify agent behavior without accessing your systems. Pure mathematical proof.
No more "trust us" - customers get verifiable evidence of your agent's performance
One Line of Code
Add verification to any function with a single decorator
Memory & Reasoning Attestation
Add verification decorators to any function. Customers get verifiable attestations of what happened.
# agent.py
from treeship import attest_memory, attest_reasoning
@attest_memory # ← Proves memory changes
def learn_user_preferences(user_id, data):
self.memory.update(user_id, data)
@attest_reasoning # ← Proves decision logic
def make_decision(context):
reasoning = self.analyze(context)
return self.decide(reasoning)
# Customer gets proof automaticallyCustomer Verification
Your customers can independently verify your agent's memory, reasoning, and performance using their own verification system.
# customer_verify.py
from treeship import verify
# Get proofs from agent
memory_proof = agent.get_memory_proof()
reasoning_proof = agent.get_reasoning_proof()
# Verify independently
verify.memory_updates(memory_proof) # ✓
verify.reasoning_chain(reasoning_proof) # ✓
# Verifiable attestationsStart Attesting in 30 Seconds
Install the SDK, create an attestation, share the verification URL. It's that simple.
pip install treeship-sdk
npm install -g treeship-cli