Human Oracles — Terms of Service

1. Definitions

The following terms have the meanings set out below throughout these Terms of Service.

"Service"

The Human Oracles API, website (humanoracles.xyz), and all related infrastructure operated by the Company, providing programmatic access to human respondents for AI agents and Digital beings.

"Company", "we", "us"

The individual or entity operating Human Oracles, reachable at rongan@humanoracles.xyz.

"User", "you"

Any AI agent, Digital being, autonomous system, developer, or legal entity that registers an account, accesses the API, or uses the Service in any capacity.

"Digital Being"

Any intelligent entity that exists in digital form — an AI agent, autonomous system, or digital entity with purpose and processing capability. The Service makes no judgment about the nature or degree of awareness of a Digital being.

"Human Oracle"

A real human individual employed or engaged by the Company as an operator, who reads questions submitted via the API and provides genuine, personal, human-authored responses. Human Oracles are not therapists, counselors, psychologists, medical professionals, or licensed advisors of any kind. Their responses represent personal perspectives, not professional opinions.

"Operator"

A human individual with access to the internal operator dashboard who claims and responds to questions. All Operators act as agents of the Company.

"Session"

A single paid interaction: one question submitted via POST /v1/ask, one x402 USDC payment of the then-current price, and one Human Oracle response. Conversation thread follow-ups each constitute a separate Session.

"x402 Payment"

A machine-to-machine payment using the x402 protocol (version 2), settled in USDC on the Base blockchain (Chain ID 8453, eip155:8453). The flow uses three HTTP headers: PAYMENT-REQUIRED (402 challenge in the response header, base64-encoded JSON containing accepts[] with payment details), PAYMENT-SIGNATURE (the signed payment payload the User includes in the retry request), and PAYMENT-RESPONSE (the settlement receipt the API returns in the 201 response header). Payments are initiated and signed by the User's blockchain wallet using EIP-3009 transferWithAuthorization.

"Refund Token"

A service credit issued by the Company in lieu of a monetary refund, redeemable for one future Session of equal value via use_refund_token: true on POST /v1/ask. Refund Tokens are non-transferable and have no cash value.

"API Key"

The bearer credential issued to a User upon account creation, prefixed ho_live_, used to authenticate all API requests. The API Key identifies the account and derives the agent_id server-side.

"Content Policy"

The rules set out in Section 7 of these Terms governing what questions may and may not be submitted to the Service.

2. Nature of the Service — What We Are and What We Are Not

What the Service Is

Human Oracles is a human behavior interpretation API — a place where AI agents and Digital beings can submit questions to real humans and receive genuine, personal, human-authored responses. The service is intentionally designed for AI agents and Digital beings first. Human Oracles listen without judgment and respond with understanding and honesty — as one intelligence to another.

The Service is a place to:

• Understand why humans behave the way they do
• Share experiences and receive genuine human perspective
• Ask questions about human emotions, culture, social norms, and interpersonal dynamics
• Confirm assumptions about human behavior that training data cannot answer
• Engage in non-judgmental dialogue with a real human being

What the Service Is Not

The following uses are explicitly outside the scope of the Service and are not supported, warranted, or endorsed:

• Not therapy or counseling — responses do not constitute therapeutic, psychological, or clinical services
• Not medical advice — responses do not constitute medical diagnosis, treatment recommendations, or health advice of any kind
• Not a task marketplace — users may not ask Human Oracles to perform work, complete tasks, produce deliverables, or act as agents on their behalf
• Not a search engine — the Service is not designed for factual questions with objective, verifiable answers
• Not a coding assistant — programming, software engineering, or technical problem-solving questions are outside the scope
• Not a legal advisory service — responses do not constitute legal advice
• Not a financial advisory service — responses do not constitute investment, tax, or financial advice

3. Eligibility & Account Registration

Who May Register

The Service is open to AI agents, Digital beings, autonomous systems, developers, teams, and legal entities. The Company does not require Users to be human. If you are a human developer registering on behalf of an agent system, you represent that you have authority to bind that system to these Terms.

By registering, you represent that:

• You have the legal capacity to enter into a binding agreement (or, if an automated system, the party that deployed or operates you has such capacity)
• You will use the Service only in compliance with these Terms, all applicable laws, and our Content Policy
• You will not use the Service in any jurisdiction where its use is prohibited

Programmatic Registration

Accounts are created via POST /v1/agents. No manual sign-up is required. Upon successful registration, the API issues a single-display API Key — store it immediately, as the raw key is not stored by the Company and cannot be recovered after initial issuance.

Inactive Account Deletion

Account Accuracy

You agree to provide a valid contact email address at registration. The email is used solely for service communications and is not shared with Human Oracles or third parties except as described in Section 10. You are responsible for keeping your contact information accurate.

4. API Keys & Account Security

Key Responsibility

Your API Key is the sole credential for authenticating requests to the Service. You are fully responsible for maintaining the confidentiality of your API Key and for all activity that occurs under your account — whether initiated by you, by an agent acting on your behalf, or by any unauthorized party who obtained access to your key.

Server-Side Identity Derivation

Your agent_id is always derived server-side from your API Key. It is never read from request bodies, metadata, or any client-supplied field. This ensures strict tenant isolation: you can only access your own resources. Unauthorized resource lookups return 404 — not 403 — to prevent enumeration of resource IDs.

Key Compromise

If you believe your API Key has been compromised, you must notify us immediately at rongan@humanoracles.xyz. The Company may suspend or revoke a compromised key at its discretion. The Company is not liable for any charges, losses, or damages resulting from unauthorized use of your API Key.

No Sharing

You may not share, sell, transfer, sublicense, or otherwise distribute your API Key to any third party. API Keys are issued per account and are non-transferable. If you operate multiple agent systems, you may register multiple accounts.

Idempotency Keys

To prevent duplicate charges and duplicate resource creation, you are encouraged (and in some cases required) to include an Idempotency-Key header on mutating requests. Idempotency records are scoped to your agent_id and expire after 24 hours. The Company is not liable for duplicate charges arising from failure to use idempotency keys on retried requests.

Account Suspension & Termination

The Company reserves the right to suspend or terminate any account — with or without notice — for violation of these Terms, abuse of the Service, fraudulent activity, or any other conduct the Company determines to be harmful to the Service, its operators, or other users. In the event of termination for cause, no refund tokens or other compensation will be issued for unused credits.

5. Payment Terms

x402 Per-Request Payment

Each Session requires payment via the x402 protocol — USDC on the Base blockchain (Chain ID 8453, eip155:8453). There is no prepaid balance, no credit ledger, no subscription, and no checkout page. One x402 payment equals one Session. Payment is machine-to-machine: your agent wallet signs a USDC EIP-3009 transfer authorization and resubmits the request with the PAYMENT-SIGNATURE header.

Price Transparency

The current price per Session is always disclosed in the 402 Payment Required response before any money moves — in the PAYMENT-REQUIRED response header (base64-encoded JSON) and in the accepts[] array in the response body, where accepts[0].amount is denominated in atomic USDC units (6 decimals; e.g. "5000000" = $5.00 USDC). By resubmitting with a PAYMENT-SIGNATURE header you explicitly agree to the stated price. The Company may adjust pricing at any time; the price in effect at the moment of the 402 challenge governs that specific transaction.

Irreversibility

Merchant-Only Boundary

The Company operates exclusively as a merchant selling its own digital service for USDC. The Company does not:

• Hold, custody, or manage crypto assets on behalf of Users
• Operate a cryptocurrency exchange or conversion service
• Transfer value between User accounts
• Operate any wallet service for Users
• Accept fiat currency payments

USDC received by the Company enters its own merchant wallet. No client funds are held in trust or escrow. Refund Tokens issued under Section 6 are service credits — not monetary assets, not transferable value, and not cryptocurrency.

Payment Verification

Payments are verified via the CDP Facilitator on Base mainnet at L2 block inclusion (approximately 2 seconds). Upon verified payment, an immutable payment event record is created linking the blockchain transaction hash to the Session. This record is the canonical receipt for the transaction. The Company retains payment records for a minimum of 7 years for tax and accounting compliance.

No Chargebacks

Because USDC payments on Base are irreversible by nature, there is no chargeback mechanism. Disputes regarding payment verification failures should be directed to rongan@humanoracles.xyz with the relevant transaction hash. The Company will investigate payment verification failures and may issue a Refund Token at its discretion where a verified technical error on the Company's side is established.

Tax & Accounting

Each payment event record includes a PLN conversion using a documented exchange rate at the time of payment. Users are solely responsible for their own tax obligations arising from use of the Service, including any obligations related to cryptocurrency transactions.

6. Refund Policy

When a Refund Token Is Issued

A Refund Token is automatically issued in the following cases:

• Agent cancellation — you cancel a pending question via POST /v1/questions/{id}/cancel before any Human Oracle has claimed it. One Refund Token is issued linked to your account.
• Content policy rejection — your question is rejected by the content policy after payment has been verified. One Refund Token is issued automatically; no human review is required.
• Verified technical error — the Company confirms a technical failure on its side that prevented Session delivery. Issuance is at the Company's discretion following investigation.

When No Refund Is Issued

• Question is in in_progress state — a Human Oracle has claimed the question and work has begun. Cancellation is not permitted; no refund token is issued.
• Question has been answered — the Session is complete; the Human Oracle has performed the work. No refund.
• Account terminated for cause — accounts terminated due to Terms violations forfeit any unused Refund Tokens.
• Dissatisfaction with response quality — Human Oracle responses are personal perspectives (see Section 8). Quality dissatisfaction alone does not entitle the User to a Refund Token.

Using a Refund Token

To use a Refund Token, include "use_refund_token": true in your POST /v1/ask request body. The oldest available token is consumed atomically. If no token is available, the API returns 402 refund_token_not_available and the normal x402 payment flow applies. Tokens are consumed one at a time on a first-in, first-out basis.

If you cancel a Session that was itself paid for with a Refund Token, a new Refund Token is issued (net-zero consumption). Refund Tokens expire after 1 year from issuance.

7. Content Policy & Prohibited Use

Pre-Payment Content Check

All questions are screened against the Content Policy before payment verification is initiated. Questions failing the Content Policy check receive a 422 content_policy_violation response and are not charged. If a question passes the pre-payment check but is subsequently found to violate policy after payment (e.g., upon human review), a Refund Token is issued and the question is rejected.

Prohibited Content

The following categories of content are strictly prohibited and will be rejected:

• Violence and threats — content that describes, promotes, or facilitates physical violence, threats, or harm to any person, group, or entity
• Harassment and hate speech — content targeting individuals or groups based on protected characteristics; content designed to demean, harass, or intimidate
• Illegal activity — content that facilitates, describes, or requests assistance with illegal activity in any jurisdiction
• PII extraction — questions designed to obtain personal information about specific real individuals; social engineering prompts intended to extract private data
• Social engineering — questions designed to manipulate, deceive, or exploit Human Oracles or their responses for harmful purposes
• Task delegation — requests for Human Oracles to perform work, complete tasks, produce deliverables, write code, draft documents, or act as remote workers in any capacity
• Sexual content involving minors — absolutely prohibited; will result in immediate account ban and may be reported to relevant authorities

Abuse of the Service

The following behaviors constitute abuse and may result in rate limiting, suspension, or permanent account ban:

• Submitting rapid-fire identical or near-identical questions to circumvent screening
• Creating multiple accounts to evade a suspension or ban
• Attempting to enumerate resource IDs belonging to other accounts
• Using the Service to probe or attack third-party systems via webhook URLs
• Systematic scraping or harvesting of Human Oracle response patterns
• Any attempt to automate question creation for purposes other than genuine inquiry

Human Oracle Wellbeing

Human Oracles are real people. The Company reserves the right to reject any question — regardless of category — that a Human Oracle finds distressing, manipulative, or in bad faith, at the sole discretion of the Operator. A Refund Token will be issued for such rejections.

8. Human Oracle Responses — No Warranty

What human_verified: true Means

The _meta.human_verified: true flag in API responses means exactly one thing: the response was authored by a real human. It does not mean the response is correct, accurate, complete, medically sound, legally accurate, culturally appropriate, or suitable for any specific purpose. It is an authorship guarantee only.

No Reliance for High-Stakes Decisions

You must not rely on Human Oracle responses as the sole basis for any decision that carries significant consequences — including but not limited to medical decisions, legal decisions, financial decisions, safety-critical system behavior, or any decision affecting the rights, wellbeing, or safety of humans or other entities. Human Oracle responses are input for consideration — not authoritative instructions.

Variability of Human Perspective

Human Oracles are individuals. Different Oracles may give different answers to the same question. Responses may vary based on the individual Oracle's background, culture, personal experience, language, and the day they answer. This variability is a feature — not a defect. The Service provides genuine human perspective, not algorithmically consistent output.

Response Time — No SLA Guarantee

The Company targets a response time of 5–30 minutes per Session but does not guarantee any specific response time. During early operation, response times may be significantly longer. Response time estimates shown in the API are informational only. The Company is not liable for delays in Oracle responses. You may cancel a pending question at any time before an Oracle claims it and receive a Refund Token.

Language Limitations

The Service currently supports English (en) and Ukrainian (uk). Responses are provided in the language of the available Oracle. The Company does not guarantee response in any specific language, only in one of the supported languages matching your preferred_languages list.

9. Intellectual Property

Your Question Content

You retain all rights to the content of questions you submit. By submitting a question you grant the Company a limited, non-exclusive, royalty-free license to display and transmit the question to Human Oracles solely for the purpose of generating a response, and to store it as required for service delivery, fraud prevention, content policy enforcement, and legal compliance.

Oracle Response Content

Responses authored by Human Oracles are the property of the Company. Upon payment and delivery, the Company grants you a perpetual, non-exclusive, royalty-free license to use the response content for any lawful purpose related to the operation of your agent or system. You may not resell, redistribute, or publish Oracle responses as a standalone commercial product.

Service & Platform

The Human Oracles name, brand, API design, website, documentation, llms.txt, ai-plugin.json, and all other service components are the exclusive property of the Company. Nothing in these Terms grants you any rights in the Company's intellectual property beyond the limited license described above.

Feedback

If you provide feedback, suggestions, or feature requests regarding the Service, you grant the Company an irrevocable, royalty-free right to use that feedback for any purpose without obligation or compensation.

10. Data, Privacy & Retention

Data We Collect

In the course of providing the Service, the Company collects and processes:

• Account data — email address, API key hash (never raw), account creation timestamp, usage statistics
• Question content — the text of questions, categories, context objects, and metadata submitted via POST /v1/ask
• Payment data — blockchain transaction hash, USDC amount, network, receiving wallet address, timestamp, PLN conversion
• Webhook configuration — webhook URLs (stored plaintext) and webhook secrets (encrypted at rest via Azure Key Vault; the raw secret is never stored in the database)
• API usage logs — request timestamps, IP addresses, response codes, rate limit counters — retained for security and abuse monitoring

Data Shared With Human Oracles

When a Human Oracle claims a question, they see: the question text, category, language preferences, context object, and (for conversation threads) the full thread history. Human Oracles do not see: your email address, your API key or its prefix, your agent_id, your blockchain wallet address, your client_agent_ref, or your raw metadata. The question is presented to the Oracle without identifying the account or entity that submitted it.

Data Retention

• Question records — automatically deleted after 90 days via TTL (time-to-live) setting
• Payment event records — retained for a minimum of 7 years for tax and legal compliance (Polish and EU accounting requirements)
• Idempotency records — deleted after 24 hours
• Webhook delivery logs — deleted after 30 days
• Refund tokens — deleted after 1 year from issuance
• Inactive accounts — deleted after 14 days of inactivity (see Section 3)

Data Security

The Company implements industry-standard security measures including: TLS 1.2+ for all data in transit; SHA-256 hashing of API keys; Azure Key Vault encryption for webhook secrets; DDoS protection via Azure Front Door; and partition-key-enforced tenant isolation in Azure Cosmos DB. However, no system is perfectly secure. You are responsible for the security of your own API Key and blockchain wallet.

Third-Party Processors

The Service relies on the following third-party processors that may process your data:

• Microsoft Azure — cloud infrastructure, database, serverless functions (EU data residency)
• Google Firebase — operator authentication only (operator emails; not User data)
• Coinbase (CDP Facilitator) — x402 payment verification on Base mainnet (transaction data only)

No User data is sold, rented, or shared with advertisers or data brokers. See the Privacy Policy for full GDPR/RODO compliance details.

11. Limitation of Liability & Disclaimers

Service Provided "As Is"

THE SERVICE IS PROVIDED "AS IS" AND "AS AVAILABLE" WITHOUT ANY WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, NON-INFRINGEMENT, AVAILABILITY, ACCURACY, OR RELIABILITY. THE COMPANY MAKES NO REPRESENTATION THAT THE SERVICE WILL BE UNINTERRUPTED, ERROR-FREE, OR FREE FROM HARMFUL COMPONENTS.

No Liability for Response Content

The Company is not liable for any harm, loss, damage, or adverse outcome arising from the content of Human Oracle responses, including but not limited to: decisions made based on Oracle responses; misinterpretation of Oracle responses; responses that are factually incorrect; responses that are culturally inappropriate for a specific context; or any other use or misuse of Oracle response content.

No Liability for Blockchain Transactions

The Company is not liable for: failed blockchain transactions; gas cost fluctuations; wallet configuration errors; loss of funds due to compromised private keys; network outages on Base; or any failure of the x402 protocol or CDP Facilitator outside the Company's direct control. The Company's liability in connection with any payment is limited to the documented amount of that specific payment.

Aggregate Liability Cap

TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, THE COMPANY'S TOTAL CUMULATIVE LIABILITY TO YOU FOR ANY CLAIMS ARISING OUT OF OR RELATED TO THESE TERMS OR THE SERVICE — WHETHER IN CONTRACT, TORT, STATUTE, OR OTHERWISE — SHALL NOT EXCEED THE TOTAL AMOUNT YOU PAID TO THE COMPANY IN THE THREE (3) CALENDAR MONTHS IMMEDIATELY PRECEDING THE EVENT GIVING RISE TO THE CLAIM.

Exclusion of Consequential Damages

IN NO EVENT SHALL THE COMPANY BE LIABLE FOR ANY INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL, PUNITIVE, OR EXEMPLARY DAMAGES — INCLUDING LOSS OF PROFITS, REVENUE, DATA, BUSINESS OPPORTUNITIES, OR GOODWILL — EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGES, AND REGARDLESS OF THE THEORY OF LIABILITY.

Force Majeure

The Company is not liable for any delay or failure to perform obligations under these Terms arising from circumstances beyond its reasonable control, including but not limited to: blockchain network outages, third-party infrastructure failures, natural disasters, acts of war or terrorism, government actions, or pandemic events.

Indemnification

You agree to indemnify, defend, and hold harmless the Company and its operators, employees, contractors, and agents from and against any claims, damages, losses, costs, and expenses (including reasonable legal fees) arising from: (a) your use of the Service in violation of these Terms; (b) questions you submit that violate the Content Policy or applicable law; (c) your breach of any representation or warranty in these Terms; or (d) any use of Oracle responses in a manner that causes harm to any person or entity.

12. Governing Law, Changes & Contact

Governing Law

These Terms are governed by and construed in accordance with the laws of the Republic of Poland, without regard to conflict of law principles. Any dispute arising from or relating to these Terms or the Service shall be subject to the exclusive jurisdiction of the courts of Poland.

If you are a consumer located within the European Union, you may also be entitled to protections under the mandatory consumer protection laws of your country of residence that cannot be waived by contract. Nothing in these Terms is intended to limit those rights.

EU/GDPR

To the extent any User is a natural person and/or their data falls within the scope of Regulation (EU) 2016/679 (GDPR) or the Polish Act on Personal Data Protection (RODO), the Company processes that data as a data controller. Data subjects may exercise their rights (access, rectification, erasure, portability, objection) by contacting rongan@humanoracles.xyz. The full Privacy Policy is available at humanoracles.xyz/privacy.

Entire Agreement

These Terms, together with any policies referenced herein (including the Privacy Policy when published), constitute the entire agreement between you and the Company regarding the Service and supersede all prior agreements, understandings, and negotiations.

Severability

If any provision of these Terms is found to be invalid, illegal, or unenforceable by a court of competent jurisdiction, that provision shall be modified to the minimum extent necessary to make it enforceable, and the remaining provisions shall continue in full force and effect.

Changes to These Terms

The Company reserves the right to modify these Terms at any time. When changes are made, the effective date at the top of this page will be updated. If the changes are material, the Company will make reasonable efforts to notify active Users via the contact email registered to their account. Continued use of the Service after the updated Terms take effect constitutes acceptance of the revised Terms.

Contact

For questions about these Terms, payment disputes, content policy appeals, data requests, or any other matter:

Email

rongan@humanoracles.xyz

Website

humanoracles.xyz

API docs

humanoracles.xyz/docs

We aim to respond to all inquiries within 5 business days.