Integration Examples
Working code showing how to integrate AttestSeal into AI agents, payment flows, and applications.
Setup
Python SDK: pip install httpx (SDK is included in examples, will be on PyPI at launch)
Or just use the API directly: GET https://api.attestseal.com/v1/check/{domain}
1. Basic Trust Check with Confidence-Aware Policy
BeginnerThe simplest integration. Check a domain before making a payment. Branch on recommended_action, which combines recommendation + confidence + cautionReason into one decision hint so your policy is the same as every other AttestSeal consumer.
from attestseal import check
result = check("stripe.com")
# One property, six policy outcomes:
if result.recommended_action == "refuse_critical": # malware/phishing detected
block_and_alert()
elif result.recommended_action == "refuse": # DENY tier
refuse_payment(result.reasoning)
elif result.recommended_action == "confirm_low_value": # CAUTION with low confidence
if amount < 25: proceed() # evidence incomplete, not necessarily bad
else: ask_user()
elif result.recommended_action == "confirm_new_domain": # CAUTION, under 1 year old
ask_user("Site is under 1 year old. Proceed?")
elif result.recommended_action == "confirm_caution": # CAUTION, weak signals
ask_user(result.reasoning)
else: # "proceed"
proceed()
# Or if you want a human-readable companion for logs:
print(result.action_message) # e.g. "Safe to proceed with this merchant."
# And the raw fields for anything custom:
print(result.trust_score) # 83
print(result.confidence) # "high"
print(result.caution_reason) # None (only set on CAUTION)
print(result.signals.ssl.score) # 100
result = check("stripe.com")
# One property, six policy outcomes:
if result.recommended_action == "refuse_critical": # malware/phishing detected
block_and_alert()
elif result.recommended_action == "refuse": # DENY tier
refuse_payment(result.reasoning)
elif result.recommended_action == "confirm_low_value": # CAUTION with low confidence
if amount < 25: proceed() # evidence incomplete, not necessarily bad
else: ask_user()
elif result.recommended_action == "confirm_new_domain": # CAUTION, under 1 year old
ask_user("Site is under 1 year old. Proceed?")
elif result.recommended_action == "confirm_caution": # CAUTION, weak signals
ask_user(result.reasoning)
else: # "proceed"
proceed()
# Or if you want a human-readable companion for logs:
print(result.action_message) # e.g. "Safe to proceed with this merchant."
# And the raw fields for anything custom:
print(result.trust_score) # 83
print(result.confidence) # "high"
print(result.caution_reason) # None (only set on CAUTION)
print(result.signals.ssl.score) # 100
Output: stripe.com (PROCEED, high confidence)
--- Attempting to pay $49.99 to stripe.com ---
Domain: stripe.com
Trust Score: 83/100
Recommendation: PROCEED
Confidence: high
Scoring Model: attestseal-v1.5.1-weights
Signals:
Reputation: 93/100 Identity: 55/55
Content: 95/100 Domain Age: 100/100
SSL/TLS: 100/100 DNS: 60/100
ACTION: Safe to proceed with this merchant.
Payment of $49.99 authorized.
Domain: stripe.com
Trust Score: 83/100
Recommendation: PROCEED
Confidence: high
Scoring Model: attestseal-v1.5.1-weights
Signals:
Reputation: 93/100 Identity: 55/55
Content: 95/100 Domain Age: 100/100
SSL/TLS: 100/100 DNS: 60/100
ACTION: Safe to proceed with this merchant.
Payment of $49.99 authorized.
Output: CAUTION with low confidence (crawl blocked by CF Enterprise)
--- Attempting to pay $8.00 to example-store.com ---
Trust Score: 58/100
Recommendation: CAUTION
Confidence: low
Caution Reason: incomplete_evidence
ACTION: Evidence incomplete. Not necessarily bad. Low-dollar OK. Confirm larger amounts.
Amount is $8 (under $25 threshold). Payment authorized.
Trust Score: 58/100
Recommendation: CAUTION
Confidence: low
Caution Reason: incomplete_evidence
ACTION: Evidence incomplete. Not necessarily bad. Low-dollar OK. Confirm larger amounts.
Amount is $8 (under $25 threshold). Payment authorized.
Output: fake domain (refuse_critical)
--- Attempting to pay $899.00 to totally-fake-xyz123.com ---
Trust Score: 23/100
Recommendation: DENY
Confidence: high
Flags: NO_IDENTITY, NO_SSL
ACTION: Refuse this transaction.
Trust Score: 23/100
Recommendation: DENY
Confidence: high
Flags: NO_IDENTITY, NO_SSL
ACTION: Refuse this transaction.
2. LangChain Agent
IntermediateAn AI agent that autonomously verifies merchants before purchasing. The agent decides on its own when to call the trust verification tool. Works with any LangChain-compatible LLM.
from langchain_openai import ChatOpenAI
from attestseal.integrations.langchain import OTTVerifyTool
# Add the trust verification tool to your agent
llm = ChatOpenAI(model="gpt-4o")
tools = [OTTVerifyTool()]
agent = create_react_agent(llm, tools)
# The agent now checks merchants automatically
agent.invoke("Buy a USB-C cable from amazon.com for under $15")
from attestseal.integrations.langchain import OTTVerifyTool
# Add the trust verification tool to your agent
llm = ChatOpenAI(model="gpt-4o")
tools = [OTTVerifyTool()]
agent = create_react_agent(llm, tools)
# The agent now checks merchants automatically
agent.invoke("Buy a USB-C cable from amazon.com for under $15")
Agent simulation output
TASK: Buy a USB-C cable from amazon.com for under $15
Agent: I need to buy from amazon.com.
Agent: Let me verify this merchant first...
[ATS Tool Response]
amazon.com: Score 67/100 (CAUTION)
Established domain, valid SSL, clean reputation.
Agent: This merchant scored 67 (CAUTION).
Agent: Amount $12.99 is within my CAUTION limit.
Agent: Proceeding with purchase at amazon.com.
TASK: Order a laptop from totally-fake-xyz123.com
[ATS Tool Response]
totally-fake-xyz123.com: Score 23/100 (DENY)
Flags: NO_IDENTITY, NO_SSL
Agent: This merchant failed trust verification (score: 23).
Agent: I cannot proceed. Searching for alternatives...
Agent: I need to buy from amazon.com.
Agent: Let me verify this merchant first...
[ATS Tool Response]
amazon.com: Score 67/100 (CAUTION)
Established domain, valid SSL, clean reputation.
Agent: This merchant scored 67 (CAUTION).
Agent: Amount $12.99 is within my CAUTION limit.
Agent: Proceeding with purchase at amazon.com.
TASK: Order a laptop from totally-fake-xyz123.com
[ATS Tool Response]
totally-fake-xyz123.com: Score 23/100 (DENY)
Flags: NO_IDENTITY, NO_SSL
Agent: This merchant failed trust verification (score: 23).
Agent: I cannot proceed. Searching for alternatives...
3. Multi-Merchant Comparison
IntermediateAgent compares prices across multiple merchants, filters out untrustworthy ones, and recommends the best deal among verified options. The cheapest option ($9.99) gets rejected because it fails trust verification.
from attestseal import OTTClient
client = OTTClient()
merchants = ["amazon.com", "ebay.com", "scosi.com",
"totally-fake-xyz123.com", "bestbuy.com"]
for domain in merchants:
result = client.check(domain)
if result.is_blocked:
print(f"REJECTED: {domain}")
else:
trusted.append(domain)
client = OTTClient()
merchants = ["amazon.com", "ebay.com", "scosi.com",
"totally-fake-xyz123.com", "bestbuy.com"]
for domain in merchants:
result = client.check(domain)
if result.is_blocked:
print(f"REJECTED: {domain}")
else:
trusted.append(domain)
Output
Agent: Finding the best deal for a Wireless Mouse...
Domain Price Score Rec Status
----------------------------------------------------------------------
amazon.com $ 29.99 67/100 CAUTION CAUTION
ebay.com $ 24.99 78/100 PROCEED TRUSTED
scosi.com $ 27.50 79/100 PROCEED TRUSTED
totally-fake-xyz123.com $ 9.99 23/100 DENY REJECTED
bestbuy.com $ 31.99 55/100 CAUTION CAUTION
Filtered out 1 untrustworthy merchant:
totally-fake-xyz123.com (score 23, $9.99)
Best trusted deal: ebay.com at $24.99
Trust score: 78/100 (PROCEED)
Domain Price Score Rec Status
----------------------------------------------------------------------
amazon.com $ 29.99 67/100 CAUTION CAUTION
ebay.com $ 24.99 78/100 PROCEED TRUSTED
scosi.com $ 27.50 79/100 PROCEED TRUSTED
totally-fake-xyz123.com $ 9.99 23/100 DENY REJECTED
bestbuy.com $ 31.99 55/100 CAUTION CAUTION
Filtered out 1 untrustworthy merchant:
totally-fake-xyz123.com (score 23, $9.99)
Best trusted deal: ebay.com at $24.99
Trust score: 78/100 (PROCEED)
4. Async Batch Checking
AdvancedCheck multiple domains concurrently. Useful for verifying a portfolio of merchants, building trust dashboards, or pre-screening a list of sites.
import asyncio
from attestseal import OTTClient
async def batch_check():
client = OTTClient()
domains = ["stripe.com", "google.com", "amazon.com",
"github.com", "cloudflare.com"]
results = await client.async_check_multiple(domains)
for r in sorted(results, key=lambda x: -x.trust_score):
print(f"{r.domain}: {r.trust_score} ({r.recommendation})")
asyncio.run(batch_check())
from attestseal import OTTClient
async def batch_check():
client = OTTClient()
domains = ["stripe.com", "google.com", "amazon.com",
"github.com", "cloudflare.com"]
results = await client.async_check_multiple(domains)
for r in sorted(results, key=lambda x: -x.trust_score):
print(f"{r.domain}: {r.trust_score} ({r.recommendation})")
asyncio.run(batch_check())
Output (10 domains in 0.2s)
Checking 10 domains concurrently...
Domain Score Rec Category
----------------------------------------------------------------------
stripe.com 83/100 PROCEED infrastructure
github.com 81/100 PROCEED infrastructure
cloudflare.com 79/100 PROCEED infrastructure
scosi.com 79/100 PROCEED consumer
apple.com 79/100 PROCEED consumer
ebay.com 78/100 PROCEED api_service
microsoft.com 77/100 PROCEED consumer
shopify.com 75/100 PROCEED api_service
google.com 73/100 CAUTION api_service
amazon.com 67/100 CAUTION consumer
Completed in 0.2s (0.0s per domain)
Summary: 8 PROCEED, 2 CAUTION, 0 DENY
Domain Score Rec Category
----------------------------------------------------------------------
stripe.com 83/100 PROCEED infrastructure
github.com 81/100 PROCEED infrastructure
cloudflare.com 79/100 PROCEED infrastructure
scosi.com 79/100 PROCEED consumer
apple.com 79/100 PROCEED consumer
ebay.com 78/100 PROCEED api_service
microsoft.com 77/100 PROCEED consumer
shopify.com 75/100 PROCEED api_service
google.com 73/100 CAUTION api_service
amazon.com 67/100 CAUTION consumer
Completed in 0.2s (0.0s per domain)
Summary: 8 PROCEED, 2 CAUTION, 0 DENY
5. Raw HTTP (Any Language)
Any LanguageNo SDK needed. The API is standard REST. Use curl, fetch, requests, or any HTTP client in any language.
# curl
curl https://api.attestseal.com/v1/check/merchant.com
# JavaScript (fetch, works in browser + Node 18+)
const r = await fetch('https://api.attestseal.com/v1/check/merchant.com');
const data = await r.json();
if (data.recommendation === 'DENY') throw new Error(data.reasoning);
if (data.recommendation === 'CAUTION' && data.confidence === 'low') {
// fetch gap, not necessarily bad; safe for small-dollar
}
# Go
resp, _ := http.Get("https://api.attestseal.com/v1/check/merchant.com")
# Ruby
response = Net::HTTP.get(URI("https://api.attestseal.com/v1/check/merchant.com"))
curl https://api.attestseal.com/v1/check/merchant.com
# JavaScript (fetch, works in browser + Node 18+)
const r = await fetch('https://api.attestseal.com/v1/check/merchant.com');
const data = await r.json();
if (data.recommendation === 'DENY') throw new Error(data.reasoning);
if (data.recommendation === 'CAUTION' && data.confidence === 'low') {
// fetch gap, not necessarily bad; safe for small-dollar
}
# Go
resp, _ := http.Get("https://api.attestseal.com/v1/check/merchant.com")
# Ruby
response = Net::HTTP.get(URI("https://api.attestseal.com/v1/check/merchant.com"))
JSON Response
{
"domain": "merchant.com",
"trustScore": 81,
"recommendation": "PROCEED",
"confidence": "high",
"cautionReason": null, // set only on CAUTION; one of: incomplete_evidence, weak_signals, new_domain, infrastructure
"scoringModel": "attestseal-v1.5.1-weights",
"reasoning": "Established domain, valid SSL...",
"brandTier": "well_known",
"siteCategory": "consumer",
"jurisdiction": { "country": "US", "crossBorderRisk": "standard" },
"signals": { ... },
"checklist": [ ... ],
"signature": "z3FXQ...ed25519", // covers score + recommendation + confidence + cautionReason
"issuer": "did:web:attestseal.com"
}
"domain": "merchant.com",
"trustScore": 81,
"recommendation": "PROCEED",
"confidence": "high",
"cautionReason": null, // set only on CAUTION; one of: incomplete_evidence, weak_signals, new_domain, infrastructure
"scoringModel": "attestseal-v1.5.1-weights",
"reasoning": "Established domain, valid SSL...",
"brandTier": "well_known",
"siteCategory": "consumer",
"jurisdiction": { "country": "US", "crossBorderRisk": "standard" },
"signals": { ... },
"checklist": [ ... ],
"signature": "z3FXQ...ed25519", // covers score + recommendation + confidence + cautionReason
"issuer": "did:web:attestseal.com"
}
The signature covers domain, trustScore, recommendation, confidence, cautionReason, and the full signals block. Fetch our public key from https://attestseal.com/.well-known/did.json to verify any response without calling us back.