Skip to content
JevForAgents
BuildsPatternsGuideResourcesWhat is Jev
⌘K
SponsorsPricingSubmit a build
BuildsPatternsGuideResourcesWhat is JevSponsorsPricingAUP & ModerationSubmit a build
Content Policy & Safety

Acceptable Use Policy (AUP)

Effective September 23, 2026. This policy governs all user submissions, directory sponsor listings, AI-assisted integrations, and platform content moderation.

1. Scope

This Acceptable Use Policy ("AUP") applies to all visitors, builders, and advertisers using JevForAgents ("the Platform"), including directory submissions, sponsor placements, and interactive search or inquiry features. This policy should be read alongside our Terms of Service and Privacy Policy.

2. Content Standards and Prohibited Categories

We strictly prohibit submitting, linking, generating, or promoting any of the following categories of content through our platform:

  • Pornographic or Adult (NSFW) Content: Sexually explicit material, adult entertainment, or non-consensual sexual imagery.
  • Violence and Graphic Content: Depictions of severe violence, gore, self-harm, cruelty, or bodily injury.
  • Hate Speech and Harassment: Content promoting discrimination, disparagement, or hatred against individuals or groups based on race, ethnicity, religion, gender, sexual orientation, disability, or national origin.
  • Child Exploitation and Abuse (CSAM): Any child sexual abuse material or content endangering minors (Zero-Tolerance: reported immediately to authorities).
  • Non-consensual Deepfakes and Impersonation: Deceptive synthetic media, unauthorized digital likenesses, or fraudulent impersonation of real persons or organizations.
  • Intellectual Property Infringement: Content or links that infringe copyrights, trademarks, trade secrets, or patents.
  • Malware, Phishing, and Deceptive Practices: Distribution of malicious software, phishing links, deceptive scams, or tools intended to bypass security systems.

3. Platform Content Moderation Mechanism

To maintain a safe and lawful platform, JevForAgents implements a multi-tiered content moderation framework combining real-time automated scanning, human editorial review, and regular sampling audits.

3.1 Automated Pre-Publishing and Real-Time Moderation

All submitted metadata, URLs, product descriptions, and user inputs are processed through an automated safety gateway before being published or indexed:

  • Input & Prompt Filtering: User inputs and listing metadata are checked against automated keyword and intent filters.
  • Safety Model Integration: Automated content moderation is powered by industry-standard safety APIs, including the OpenAI Moderation API and automated link safety scanners, detecting violations across sexual, hate, harassment, self-harm, and violence categories.
  • Anomaly & Abuse Detection: High-frequency submissions, automated bot spam, and abnormal payment patterns are intercepted before checkout or live indexing.

3.2 Human Review & Trust and Safety Team

  • Any listing or submission flagged by automated safety filters is routed to an expedited manual review queue.
  • Community and visitor reports received via our dedicated reporting channels are investigated directly by our safety review team.
  • Reviewers evaluate destination URLs, context, and claims to ensure compliance with our content standards.

3.3 Regular Sampling and Audits

  • We conduct recurring spot-checks of published listings and third-party links to detect post-approval alterations or compromised destinations.
  • Automated filter thresholds and keyword blocklists are calibrated continuously to maintain high accuracy and reduce false positives.

3.4 Severity Levels (L1–L4)

LevelDescriptionModeration ProcessEnforcement Action
L1 CriticalCSAM, terrorism, severe physical threatImmediate automated block + manual verificationImmediate takedown, permanent ban, law enforcement report
L2 High RiskMalware, phishing, explicit adult content, non-consensual deepfakesAutomated quarantine + priority manual reviewRemoval of listing, suspension of placement, full investigation
L3 Medium RiskMisleading claims, IP infringement notice, low-quality spamHuman review queueFormal warning, modification requirement, or delisting
L4 Low RiskMinor formatting inaccuracies, broken redirect linksAutomated monitor / community reportNotice to submitter to update link or details

4. Underlying AI Models Disclosure

JevForAgents showcases and integrates agent architectures built in collaboration with mainstream, industry-leading AI foundation models:

  • Primary Foundation Models: Integrations, benchmarks, and showcased builds run in conjunction with established frontier models, including OpenAI GPT-4o / GPT-4 Turbo, Anthropic Claude 3.5 Sonnet, and Google Gemini 1.5 Pro / Flash.
  • Role of Jev: Jev serves as a specialized sub-50ms deterministic decision, routing, and guardrail layer designed to guide foundation models, evaluate tool execution schemas, and gate outputs before execution.
  • Safety Architecture: Content safety is enforced through foundational safety pipelines (such as the OpenAI Moderation API) alongside platform-level safety filters.

5. Reporting and Community Safeguards

We encourage users, rights-holders, and security researchers to report any violating content, deceptive listings, or intellectual property concerns.

  • Confidentiality: Reporter identity is protected and treated with strict confidentiality.
  • Impartial Review: Every report is evaluated objectively according to our published content standards.
  • Protection Against Abuse: Submitting repeated bad-faith or malicious reports may lead to submission restrictions.

6. Reporting Channels

To file a violation report, please email our safety desk:

  • Dedicated Safety & Abuse Email: l1137622803@gmail.com
  • Required Information: Please provide the URL of the affected listing or page, specific grounds for violation, and relevant supporting evidence or documentation.

7. Response Timelines

Severity LevelInitial Response WindowResolution & Action Target
L1 CriticalWithin 2 hoursWithin 24 hours
L2 High RiskWithin 24 hoursWithin 3 business days
L3 Medium RiskWithin 3 business daysWithin 7 business days
L4 Low RiskWithin 5 business daysWithin 15 business days

8. Appeals Process

If your submission or sponsor placement has been restricted, modified, or removed and you believe the decision was made in error, you may file an appeal:

  • Appeal Channel: Send an email to l1137622803@gmail.com within 14 days of the enforcement action.
  • Independent Review: Appeals are reviewed independently by a safety officer who was not involved in the original enforcement decision.
  • Resolution Timeframe: Appeal determinations are issued within 10 business days.

9. Policy Updates and Transparency

We periodically update this Acceptable Use Policy to reflect evolving regulatory standards, AI safety advancements, and operational needs. Changes take effect upon publication on this page. Material revisions will be highlighted with a revised effective date.

10. Contact Us

For general safety inquiries, AUP questions, or compliance clarifications, contact us at: l1137622803@gmail.com.

JevForAgents

The practical database and guide to building AI agents with Jev.

Build with Jev

  • Codex
  • Claude Code
  • Browser Use
  • MCP
  • Agent Skills
  • Build Your Own Agent

Patterns

  • Browser Agents
  • Agent Routing
  • Tool Selection
  • Agent Evaluation
  • Guardrails
  • MCP / Skills
  • Search & Semantic Retrieval
  • Realtime & Audio Agents
  • Coding Agents

Database & Research

  • All Builds
  • Resources Library
  • Jev Engineering PDF Guide
  • Video Demos
  • GitHub Projects
  • Jev Agent Statistics Coming soon
  • Evidence Protocol

Community

  • Submit a Build
  • Contact support
  • Suggest a Correction
  • Sponsors Leaderboard
  • Sponsor Pricing
  • Terms of Service
  • Acceptable Use Policy (AUP)
  • Privacy Policy
  • Official Jev Docs ↗

Independent from TypeSafe AI. What is Jev · Agent builder’s guideOnline counts browsers seen within the last 75 seconds. Views count each page load or in-site page change, including repeat visits. We use a random first-party browser ID for the online count and do not store IP addresses.