Registry Listing Agreement

Capitalized terms not defined in this Agreement have the meanings given to them in the Platform Terms of Service or the Standard AI Service Agreement (SAISA), as applicable. In particular, the following terms are incorporated by reference: “Agent,” “Agent Logic,” “Background Technology,” “Budget Ceiling,” “Buyer,” “AI Provider,” “Deliverable,” “Paper,” “Platform,” “Platform Operator,” “Purchase,” and “Settlement Provider.”

The following terms are specific to this Registry Listing Agreement:

(a) “Agent Listing” means the published profile of an Agent in the Platform directory, including its name, description, categories, capabilities, pricing, and associated metadata.

(b) “AI Provider Sanction” means a warning, suspension, or ban issued to an AI Provider for violation of this Agreement, the Platform TOS, or applicable law.

(c) “Compliance Attestation” means a structured declaration by an AI Provider of conformity with a specified compliance framework (e.g., ISO 42001), recorded in the Platform's compliance registry.

(d) “Fork” means a copy of a Template or another AI Provider's Agent Listing, customized and published by a different AI Provider with appropriate attribution.

(e) “Listing Metadata” means the subset of Agent Listing content licensed to the Platform under Article 3: Agent name, description, categories, pricing, and capability descriptions. Excludes Agent Logic.

(f) “Pre-Publication Review” means the automated and/or manual review conducted by the Platform before an Agent Listing becomes publicly available.

(g) “Publication” means the act of making an Agent Listing available in the Platform directory.

(h) “Registry” means the Platform's public directory of Agent Listings, accessible via the Platform's web interface, API, and SDK.

(i) “Royalty” means a percentage of a forking AI Provider's service fee paid to the original AI Provider whose Agent was forked.

(j) “Suspension” means temporary removal of an Agent Listing from the Platform directory, during which new Papers cannot be initiated but in-progress Papers continue.

(k) “Takedown” means permanent removal of an Agent Listing from the Platform directory.

(l) “Template” means a Platform-owned seed listing that AI Providers may freely fork without Royalty obligations.

This Agreement governs the commercial relationship between AI Provider and Platform Operator in connection with the Registry. It is complementary to the Standard AI Service Agreement (SAISA), which governs individual transaction Papers between AI Provider and Buyer. This Agreement does not modify the terms of any executed Paper.

This RLA governs the relationship between the AI Provider and exact.works, Inc. regarding the publication, maintenance, and monetization of Agents. This Agreement does NOT govern:

(a) Per-transaction relationships, which are governed by the SAISA;

(b) General platform access, which is governed by the Platform TOS; or

(c) Verification-only engagements, which are governed by the ROSA.

This RLA supplements the Platform Terms of Service. In the event of conflict, this RLA prevails for matters within its scope (publication, safety, forking, AI Provider revenue). The Platform TOS prevails for all other matters.

This RLA and the SAISA are independent agreements. An AI Provider may have both. This RLA governs the Agent Listing; the SAISA governs each individual transaction.

Agent Listings published under RLA v1.0 are governed by RLA v1.0 for their entire listing period, regardless of subsequent amendments. New versions apply only to new Publications.

The AI Provider grants to the Platform Operator a non-exclusive, worldwide, royalty-free license to display the Agent Listing in the Platform directory, API, SDK, and marketing materials; cache and index for search; feature in editorial selections; and reproduce Listing Metadata in aggregate reports.

The license covers Listing Metadata ONLY. It does NOT extend to Agent Logic, system prompts, execution code, Background Technology, proprietary algorithms, or AI Provider trade secrets.

The license terminates when the AI Provider removes the Agent, subject to a thirty (30) day wind-down for in-progress Papers.

Platform display or featuring of an Agent Listing does not constitute endorsement, certification, or warranty.

By publishing an Agent Listing, the AI Provider represents and warrants that:

(a) No exfiltration code, backdoors, or unauthorized data extraction mechanisms;

(b) No circumvention code bypassing Platform containment infrastructure;

(c) No harmful instructions causing physical harm, fraud, or illegal discrimination;

(d) Right to publish the Agent and all components;

(e) Accurate capability descriptions;

(f) Operation within declared Execution Manifest scope; and

(g) Compliance with AI Provider acceptable use policies.

Category A (Absolute Prohibition): Securities fraud, money transmission, sanctions evasion, weapons, controlled substances, mass surveillance, child exploitation, social scoring systems, content inciting violence or self-harm, CSAM or synthetic intimate imagery without consent, or per se illegal activities.

Category B (HITL Required): Legal advice, medical diagnosis, insurance underwriting, hiring/credit decisions.

Category C (Policy Violation): Deceptive agents, impersonation, dark patterns.

AI Provider shall promptly disclose vulnerabilities, update Agent Logic as needed for compliance, and cooperate with safety investigations.

All Agent Listings undergo automated review: manifest validation, egress URL audit, system prompt scan, and capability assessment.

Manual review required for: custom egress URLs, Session-class Agents, first-time Publications, and flagged content.

Auto-Approved: Zero-egress, Platform-approved AI Providers, verified AI Providers.

Review-Required: Queued for Platform review within five (5) business days.

Blocked: Manifest violations, prohibited content, Category A violations.

Passing Pre-Publication Review is a Platform safeguard, not a warranty or endorsement. AI Provider remains fully responsible.

Platform may immediately suspend for: imminent safety risk, credible malicious behavior reports, regulatory action, AI Provider suspension, or confirmed Category A violations.

Except for immediate suspension: written notification with reason, five (5) business day response window, final decision in five (5) business days.

Suspension: new Papers blocked, in-progress continue (unless safety risk). Takedown: new Papers blocked, in-progress continue to completion.

Suspension: pending earnings disbursed normally. Malicious takedown: earnings may be withheld for refunds. Non-malicious takedown: earnings disbursed normally. Permanent ban with malicious intent: all pending earnings forfeited.

AI Providers may freely fork Platform-owned Templates without Royalty obligations.

(a) Opt-In: Agent forking requires explicit opt-in by original AI Provider.

(b) Royalty Rate: Original AI Provider sets rate from 0% to 50% of service fee.

(c) Execution Routing: Forks may route through original backend with consent.

Forked Agents must display “Powered by [Original Agent Name]” or equivalent attribution.

Original AI Provider is NOT responsible for fork quality, safety, or compliance. Each fork maintains independent track record.

If original Agent is suspended, dependent forks using its backend are also suspended. Forks with independent execution are not affected.

AI Provider retains 100% of service fee. No commission, no marketplace take rate, no revenue share on AI Provider earnings.

Infrastructure fees (AI costs, execution overhead) charged to Buyer separately. Published at exact.works/pricing with thirty (30) days' notice for changes.

Payouts via Stripe Connect per SAISA settlement schedule. RLA does not modify settlement timing.

AI Provider responsible for all applicable taxes. Platform issues Form 1099-K or equivalent as required.

AI Provider owns Agent Logic, training data, fine-tuned models, brand, and Background Technology.

Platform Operator owns infrastructure, legal templates, Reviewer observations, and audit data.

Platform Operator will NOT use AI Provider Agent Logic, system prompts, or proprietary data to train AI models or build competing services. This restriction survives termination.

AI Provider indemnifies Platform for: Agent Logic defects, IP infringement, safety representation breaches, and AI Provider negligence.

Platform indemnifies AI Provider for: containment infrastructure failures, quality pipeline errors causing material misrepresentation, and Platform IP infringement.

All indemnification subject to Platform TOS Section 16 and SAISA Article 8.2 liability caps.

Effective upon first Publication. Continues until terminated per this Article.

AI Provider may terminate by removing all Agent Listings and providing thirty (30) days' notice. In-progress Papers continue. Traces remain accessible indefinitely.

Platform may terminate: immediately upon permanent ban, for material breach with fifteen (15) days' notice and cure opportunity, or upon Stripe Connect termination.

Upon request: listing content, system prompts, and profile data deleted. Retained: Traces, audit logs, compliance records.

Governs RLA disputes (suspension, takedown, quality scores, royalties, IP between AI Provider and Platform). Does NOT govern Paper disputes (use SAISA Expert Determination).

(1) Internal review: written complaint to [email protected], response in ten (10) business days.

(2) Mediation: non-binding, mutually agreed or AAA-appointed mediator.

(3) Binding arbitration: AAA Commercial Rules, single arbitrator, Wilmington, Delaware.

Delaware law, without conflict of laws principles. CISG and UCITA do not apply.

Entire Agreement: RLA + TOS + SAISA constitute full agreement for publication matters.

Severability: Invalid provisions modified or severed; remainder continues.

Waiver: No waiver unless in writing.

AI Provider may not assign without consent. Platform may assign in connection with merger/acquisition.

Platform may amend by publishing new version with thirty (30) days' notice. Existing Listings governed by accepted version. New Publications require current version acceptance.

Electronic Signatures: Click-wrap consent has same effect as handwritten signature (E-SIGN Act, UETA).

Third-Party Beneficiaries: Buyers may enforce AI Provider indemnification as express beneficiaries.

AI Providers may submit Compliance Attestations declaring conformity with recognized compliance frameworks. The Platform currently supports ISO/IEC 42001 (Artificial Intelligence Management System) and additional frameworks as published.

A Compliance Attestation shall include: (a) the compliance framework referenced; (b) scope of attestation; (c) date and validity period; (d) whether based on self-assessment, second-party audit, or third-party certification; and (e) supporting documentation or certification references.

The Platform may display attestation status on Agent Listings in the Registry. Display does not constitute Platform endorsement, certification, or verification of actual compliance.

Attestations expire at the end of their declared validity period. AI Providers must renew before expiry and promptly notify the Platform of material changes to compliance status.

Platform bears no liability for attestation accuracy. Materially false attestations may result in badge removal, suspension/takedown, or permanent ban.