Policy and Privacy

Last updated: September 11, 2026

1. Introduction

1.1. General information about the Service

EntoCode is a software suite for software development with the assistance of artificial intelligence models. It includes a desktop application, a website, and related account services. The Service is intended for work on the User's local projects: exploring a codebase, preparing and applying changes, running commands, and keeping project sessions in a single workspace. Use of EntoCode presupposes good-faith application of its tools for lawful professional or educational purposes. This document sets out the principles of information handling and the rules of interaction between the User and the Administration.

1.2. Subject and purpose of this document

This Policy and Privacy (the "Policy") governs the collection, storage, use, and protection of information connected with use of the Service. The Policy forms an integral part of the terms on which access to EntoCode is granted and applies together with any other rules of the Service. Review of the Policy and acceptance of its provisions are a condition of registration, sign-in, and further use of the Service's functions. If the User does not agree with the Policy, the User must discontinue use of the Service.

1.3. Amendment of this document

The Administration may amend this document at any time, without public notice. A new edition takes effect from the moment it is posted in the Service, or from such other date as the Administration may specify if such a date is stated. Continued use of the Service after a change to the Policy constitutes the User's acceptance of the updated edition. The Administration is under no duty to send personal notices of each amendment; it is the User's responsibility to review the current version.

1.4. Prohibition of malware development

The Administration of the site is categorically opposed to the creation of viruses. EntoCode is not intended for, and must not be treated as, a means of developing, distributing, testing, or concealing malicious code, including viruses, worms, trojans, ransomware, and other instruments of unauthorized interference. Use of the Service for such purposes is prohibited and constitutes a material breach of the rules. The Administration may immediately restrict or terminate the User's access upon detection of such activity and may disclose information to competent authorities where required by law.

1.5. Scope and other terms

This Policy applies to the website, the desktop application, the user account, and the billing and operational processes connected with access to models and a credit plan. Certain information may be processed by engaged contractors (hosting, authentication, payments, and model providers) to the extent necessary to operate the Service. Symbols and names of third parties are used for informational purposes only and do not imply that EntoCode is affiliated with those organizations. Matters not addressed in this section are resolved under the remaining provisions of the Policy and applicable law.

2. Collected data and analytics

2.1. Categories of data used by EntoCode

EntoCode uses only those categories of data that are required to operate the account, deliver the agent, and account for plan credits. Account data include the User's email address, display name, authentication tokens, and the selected plan. Product data include project records (name, workspace path, and description), session identifiers, session titles, and encrypted chat history stored so that a session can be resumed after sign-in. Billing data include the monthly credit pool, amounts spent, remaining credit, the billing period, and individual usage events tied to a model identifier. Technical data may include application version, host operating system as supplied for workspace awareness, and basic diagnostic information needed to keep sign-in and proxying secure. EntoCode does not treat the User's entire local disk as collected data: files and folders remain on the User's machine unless the User, through the agent, causes them to be read and included in a request. Model output, tool results, and user prompts that are sent in a turn become part of that request and may be retained in encrypted session history.

2.2. Purposes of processing

Account data are used to create and authenticate the User, apply the correct plan, and restore access across the website and the desktop application. Product data are used to organize chats by project, inject workspace context that the User has associated with a session, and display prior messages. Billing and usage data are used to enforce monthly credit limits, show remaining credit and history in the account, and prevent overspend. Technical and diagnostic data are used to maintain availability, investigate faults, and protect the Service from abuse. Analytics in EntoCode are operational and billing-oriented; they are not used to sell personal data to third parties or to build advertising profiles. Processing is limited to what is necessary for the corresponding purpose. Refusal to provide account data makes registration and paid access impossible. Continued use of the agent after sign-in constitutes acceptance of processing for the purposes stated in this section.

2.3. Telemetry collected by the agent

When the User sends a request through the EntoCode agent, the Service records operational telemetry required to bill the turn and to display usage. That telemetry includes the identifier of the selected model, the number of prompt tokens, the number of completion tokens, the cost of the request, the billing period, and the time of the event. The Service also maintains a monthly spend total and a daily request count for the User. Session-level metadata (session identifier, project association, and title) may be updated in connection with the same activity. The agent does not, as telemetry, upload the User's full repository; only the fragments, paths, and tool results that the agent includes in a given turn are transmitted as part of the model request. Ratings or feedback that the User submits about a model reply may be stored together with the related message and model identifier. Telemetry described in this clause is generated automatically by use of the agent and does not require a separate toggle for each request.

2.4. Where telemetry is stored

Usage telemetry is stored on the EntoCode control plane in the Service's operational databases, together with the User's account and encrypted session records. Individual usage events are retained in association with the User identifier, the model identifier, token counts, cost, and timestamp, and are aggregated into monthly spend for the current billing period. Daily request counters are stored for the same User for operational limits and accounting. Encrypted chat history is stored separately from raw usage events, as ciphertext with a nonce, so that message content is not held in clear text on the control plane. Telemetry is not stored on the User's local machine as the system of record; the local application may cache or display figures retrieved from the control plane. Access to stored telemetry is restricted to systems that provide the account, proxy, and billing functions. Retention follows the periods stated later in this Policy and any longer period required for security, dispute resolution, or applicable law. The Administration may delete or anonymize telemetry when it is no longer needed for those purposes.

2.5. Automatic consent to processing by model providers

By creating an account, signing in, or sending a request through the agent, the User automatically agrees that data necessary to generate a model response may be transmitted to, and processed by, the providers of the models available in EntoCode. The catalog of available models is determined by the User's plan and is shown in the product; it may include models associated with providers such as Google LLC, OpenAI LLC, Anthropic PBC, xAI, and other suppliers listed in the Service. The User does not grant a separate consent for each provider each time a model is selected: selection or use of a model from the available list constitutes consent to that provider's processing of the prompt, attached context, tool results, and related metadata for that request. Each provider processes such data under its own terms, policies, and retention practices, which EntoCode does not control. The Administration does not represent that a provider will refrain from using request data for its own safety, abuse, or service-improvement programs where that provider's terms so allow. If the User does not agree to processing by the providers of models listed in EntoCode, the User must not send requests through the agent and must discontinue use of those models. Names and symbols of providers appear in the Service for identification only and do not mean that EntoCode is affiliated with those companies.

3. Storage, access, deletion, and telemetry controls

3.1. Storage of User data

User data that leave the local workspace are stored on the EntoCode control plane for as long as they are needed to provide the account, the agent, and billing. Account identifiers, plan status, and authentication material are held in the Service's operational databases so that the User can sign in on the website and in the desktop application. Encrypted chat history is stored as ciphertext with a nonce, together with session and project metadata required to restore a conversation. Usage telemetry described in Section 2 is stored in the same control-plane environment as usage events, monthly spend totals, and daily request counters. Local project files remain on the User's device and are not mirrored in full to EntoCode storage. Cached copies may exist temporarily in application memory or on disk for display and retry; they are not the system of record. Storage is limited to systems that operate the Service. The Administration does not use stored User data to build advertising profiles or to sell lists of Users.

3.2. Obtaining a copy of User data

The User may request a copy of personal data that EntoCode holds about that User. Such a request may cover account details, plan and credit figures, usage history available in the account, project and session metadata, and other records that the Administration can locate and identify as relating to the requester. The User should submit the request through the support channels listed on the EntoCode website and must use the email address associated with the account so that the Administration can verify identity. The Administration will provide the available data in a structured, commonly used form where technically feasible, or in another readable form if a structured export cannot be produced. The Administration may refuse or narrow a request that is manifestly unfounded, excessive, or that would disclose secrets of other Users or of the Service. Data that exist only on the User's local machine are not in the Administration's possession and cannot be "obtained" from EntoCode; the User already controls those files. Data already transmitted to a model provider must be requested from that provider under that provider's own procedures. The Administration will respond within a reasonable period and may ask for additional information if the request is unclear.

3.3. Deletion of User data

The User may request deletion of personal data held by EntoCode, including the account, encrypted session history, project records, and stored usage events that can be isolated to that User. A deletion request should be sent through the same support channels and from the account email. Upon verification, the Administration will erase or irreversibly anonymize the data it controls, except where retention is required for security, fraud prevention, dispute resolution, tax, or other applicable law. Deletion of the account ends access to the Service; unused credits are not a ground to keep the account if the User insists on erasure, unless the law requires a billing record to be kept. Deletion on the control plane does not erase copies that a model provider may have received during a request; those copies are governed by the provider's terms. Local files, logs, and caches on the User's device must be removed by the User if the User wishes them gone. The Administration may take a short operational period to complete deletion across backups and replicas. After deletion, restoration of the same account and history cannot be guaranteed.

3.4. Disabling telemetry and related controls

Operational usage telemetry is generated when the User sends a request through the agent, because the Service must record the model, token counts, and cost in order to enforce the credit plan and to show remaining credit. That class of telemetry cannot be switched off while the User continues to use the agent, as billing and abuse controls would otherwise fail. The User may stop the generation of new agent telemetry at any time by not sending further requests and by signing out of the desktop application and the website. Optional product diagnostics that are not required for billing, if introduced, will be described in the Service and may be disabled where a control is provided; until such a control exists, the User should treat agent usage itself as the act that creates telemetry. The User may request deletion of stored telemetry as provided in clause 3.3, subject to mandatory accounting retention. Disabling or ceasing telemetry does not withdraw consent already given to a model provider for data that were transmitted in prior requests. The Administration is not obliged to offer a mode of the agent that performs model calls without recording usage. If the User does not accept these conditions, the User must discontinue use of the agent.

3.5. Limits, residual copies, and account tools

The User may update profile details available in the account and may sign out of the current browser or application session. Sign-out ends the active session token on that client; it does not by itself delete stored history or telemetry. The Administration may decline to fulfill an access or deletion request if it cannot verify the requester as the account holder. Residual copies may remain in encrypted backups for a limited rotation period and are then overwritten in the ordinary course. Legal holds, security investigations, and binding orders may delay or restrict erasure. Third-party processors and model providers are not under the Administration's exclusive control; the User may need to exercise rights directly against those parties. Nothing in this section obliges EntoCode to provide data that it does not hold or to reconstruct data that were never stored. Questions about storage, access, deletion, or telemetry should be directed to the support channels listed on the EntoCode website.

4. Compliance with the law of the Russian Federation

4.1. Applicable law and the Constitution of the Russian Federation

The Administration intends that the Service be used in a manner consistent with the Constitution of the Russian Federation and with the legislation of the Russian Federation applicable to the User's activity. The User is responsible for ensuring that prompts, code, commands, published materials, and other use of EntoCode do not violate the Constitution of the Russian Federation, federal laws, or other binding norms. The Service must not be used to call for violent change of the constitutional order, to disclose state secrets, to incite hatred, to commit fraud, or to carry out other acts prohibited by the law of the Russian Federation. The Administration does not provide legal advice and does not certify that a particular project is lawful; that assessment remains with the User. Nothing in this Policy limits the application of mandatory norms of the Russian Federation where such norms apply. If a provision of this Policy conflicts with a mandatory norm of the Russian Federation, the mandatory norm prevails for that conflict. Use of the Service after publication of this section constitutes acknowledgment of these conditions.

4.2. Claims against EntoCode and further consideration

In the event of claims, demands, complaints, or other presentations brought against EntoCode, all further consideration of such matters shall proceed strictly in accordance with the Constitution of the Russian Federation. The Administration will treat the Constitution of the Russian Federation as the supreme legal basis for assessing such claims, together with federal laws and other acts adopted pursuant to the Constitution. Pre-trial correspondence, internal review, and any subsequent procedure that the Administration conducts in respect of a claim against EntoCode shall be guided by those instruments. The User agrees that disputes arising from use of the Service, including claims against EntoCode, are to be considered under that legal framework and not under a foreign constitution. Filing a claim does not by itself suspend the Administration's right to apply this Policy, including the right to restrict access as provided below. The Administration may request documents and explanations reasonably needed to consider a claim. Consideration of a claim does not imply admission of liability. Time limits and formalities prescribed by the legislation of the Russian Federation apply to the extent they govern the relevant procedure.

4.3. Payment-system rules

Payments for plans, credits, and related charges are processed in accordance with the rules of the payment systems and payment service providers used by the Service. The User must provide accurate payment details and must not use stolen, unauthorized, or otherwise unlawful means of payment. The Administration may decline, reverse, or hold a payment where required by the payment system, the acquiring bank, anti-money-laundering rules, or the legislation of the Russian Federation. Chargebacks, refunds, and payment disputes are handled under the rules of the relevant payment system and the terms of the payment provider, in addition to this Policy. The Administration is not a payment system and does not control settlement times, currency conversion, or issuer decisions. Fees charged by the payment system or the User's bank are borne by the User unless the payment system's rules or applicable law require otherwise. The User authorizes the Administration and its payment processors to exchange the data necessary to complete, verify, or contest a payment. Violation of payment-system rules may result in refusal of service and in the measures described in clause 4.4.

4.4. Immediate disablement of an account for violation of the Constitution of the Russian Federation

The Administration may immediately disable a EntoCode account if the User's activity violates the Constitution of the Russian Federation. Disablement may include blocking sign-in, suspending the agent, revoking tokens, and withholding further use of paid features, without prior notice where delay would be inconsistent with the gravity of the violation. Grounds include, without limitation, use of the Service for acts aimed at the violent seizure of power, incitement of social, racial, national, or religious hatred, terrorist or extremist activity, or other conduct that the Administration reasonably assesses as contrary to the Constitution of the Russian Federation. The Administration may act on information available to it, including content of requests, reports from other Users, and lawful demands of competent authorities. Immediate disablement does not require a court judgment to take operational effect; a later review may confirm, modify, or lift the measure. Unused credits are not a ground to keep an account active if disablement is applied under this clause. The Administration may retain records necessary to explain the measure to competent authorities or to consider a claim as provided in clause 4.2. Re-enablement, if any, is at the Administration's discretion after the User has demonstrated that the violating activity has ceased.

5. Use of the Service by children and responsibility for harmful use of the agent

5.1. Minimum age

The Service officially permits use by children who have reached the age of fourteen (14) years. A person under fourteen years of age is not authorized to create an account, sign in, or send requests through the agent. By registering or using EntoCode, the User confirms that the User is at least fourteen years old. If the Administration learns that an account belongs to a person under fourteen, the Administration may disable the account and delete or restrict the related data as provided in this Policy. Age confirmation may be requested where the Administration has reasonable doubt. False indication of age is a material breach of this Policy. Access at fourteen years or older does not relieve the User of the duties set out in this section and in Section 1.4. Parents or other legal representatives of a User aged fourteen or older are expected to explain these rules to the User and to supervise use where the law of the Russian Federation so requires.

5.2. Use by Users aged fourteen and older

A User who is fourteen or older may use the website, the desktop application, and the agent within the limits of this Policy and applicable law. The Administration does not treat such a User as lacking capacity solely by reason of minority, but the User remains subject to restrictions imposed by the legislation of the Russian Federation on transactions and on processing of personal data. Where the law requires the consent or participation of a parent or other legal representative, that person is responsible for granting it and for the consequences of the minor's use of the Service. The Administration is not obliged to obtain a separate written consent from a parent for every request sent through the agent. A parent or legal representative may contact the Administration to request restriction or deletion of an account that belongs to a minor, subject to verification. Paid plans used by a User aged fourteen or older must be paid by a person who is legally able to enter into the payment. The Administration may refuse service if it cannot establish that the age and payment conditions are met.

5.3. Responsibility for all queries despite the prohibition of malware

Notwithstanding the fact that EntoCode condemns and categorically prohibits the creation of viral and other malicious software, the User automatically accepts that responsibility for all queries entered by the User lies with the User. The prohibition in Section 1.4 does not transfer authorship, control, or legal liability for a prompt, command, or generated result from the User to the Administration. The agent executes or transmits what the User instructs, including text, paths, and tool calls that the User initiates. The User is responsible for the content of every request, for the decision to send it, and for any use of the output, whether or not the request was made in good faith. The Administration does not review each query in advance and does not assume liability for harm caused by the User's instructions. Automatic acceptance of this allocation of responsibility occurs upon registration, sign-in, or the sending of any request through the agent. If the User does not accept that all entered queries are the User's responsibility, the User must not use the agent.

5.4. Harmful purposes and consequences

Use of the agent for harmful purposes includes, without limitation, the creation, improvement, concealment, or distribution of viruses and other malware, attacks on information systems, fraud, and other unlawful acts. Such use is forbidden even if the User is fourteen or older and even if the User claims that the activity was experimental, educational, or hypothetical. The Administration may immediately disable the account as provided in Section 4.4 and may refuse further service. Responsibility toward third parties, including compensation for damage, remains with the User who entered the queries and not with EntoCode, except where mandatory law provides otherwise. The User shall indemnify the Administration against claims arising from the User's harmful or unlawful use of the agent, to the extent permitted by law. Reports of such use may be considered without prior notice to the User where notice would impede the response. Nothing in this section is a license to commit a prohibited act. Questions about age or about this allocation of responsibility may be sent through the support channels listed on the EntoCode website.

6. Reliability and privacy of conversations with the agent

6.1. No guarantee of one-hundred-percent reliability or privacy

EntoCode does not guarantee one-hundred-percent reliability or one-hundred-percent privacy of the User's conversations with the agent. Transmission of a request depends on networks, the availability of the control plane, and the availability of third-party model services, any of which may fail, delay, or interrupt a session. A message that appears in the application may not have been delivered, may have been delivered only in part, or may have been processed by a provider after a retry. Confidentiality of a conversation cannot be absolute once the content of a turn has left the User's device. The Administration does not warrant that a conversation will remain secret from every third party, including the operator of the selected model. Interruptions, errors, or loss of a turn do not by themselves constitute a breach of this Policy. The User should not treat the agent as a channel for information that must never leave the User's machine. If the User requires a guarantee of complete privacy, the User must not send that information through the agent.

6.2. Third-party agents and servers of their developers

EntoCode uses third-party models and related agent services whose developers receive requests on their own servers. When the User sends a prompt, attached context, or tool result through the EntoCode agent, that material is forwarded to the provider of the selected model in order to generate a response. Those providers include, depending on the catalog available under the User's plan, operators associated with Google LLC, OpenAI LLC, Anthropic PBC, xAI, and other suppliers listed in the Service. The Administration does not operate those external servers and does not supervise their staff, logs, or retention. Each provider applies its own security, logging, and privacy practices, which may permit storage, review for abuse, or other processing described in that provider's terms. A copy of a request that has reached a provider's server is outside the exclusive possession of EntoCode. Names and symbols of those companies are used for identification only and do not mean affiliation. By using a model from the available list, the User accepts that the corresponding developer will receive the request on that developer's infrastructure.

6.3. Storage inside the EntoCode ecosystem

Within the EntoCode ecosystem, the Administration stores User data by the means described in Sections 2 and 3, including encrypted chat history and usage records on the control plane. Account sign-in, session lists, and credit figures are held in Service systems so that the User can resume work after leaving the application. Encryption in transit and encrypted storage of message content on the control plane are used to reduce the risk of unauthorized reading inside that ecosystem. Access to those stores is limited to components that provide the account, the proxy, and billing. Local project files that the User does not send in a turn remain on the User's device and are not copied in full into EntoCode storage. The Administration treats data that remain inside EntoCode systems as data under its operational control. That control applies only to the ecosystem described in this Policy. Reliability of storage inside EntoCode does not extend a guarantee to systems that EntoCode does not operate.

6.4. Absence of control outside EntoCode

Outside the EntoCode ecosystem, the Service does not exercise control over User data. Once a request has been delivered to a third-party model provider, the Administration cannot warrant how that provider will store, use, share, or delete the content. The Administration cannot remotely erase a copy that remains on a provider's server, nor can it audit that server on the User's demand. Payment processors, hosting contractors, and other parties engaged only for a limited function likewise operate outside the Administration's day-to-day control, except as their contracts allow. The User accepts this limit as a condition of using third-party models through EntoCode. Claims about processing that occurs solely on a provider's infrastructure must be addressed to that provider under its own terms, without prejudice to claims against EntoCode that concern the Service itself. Nothing in this section obliges EntoCode to guarantee the conduct of a third party. If the User does not accept that conversations leave EntoCode when a model is called, the User must not use the agent.

7. Compliance with supplier documents

7.1. Full prohibition of violating supplier rules

EntoCode fully prohibits any violation of the current legal documents of the companies that supply models and routing used in the Service. Those suppliers include Google LLC, Anthropic PBC, xAI, OpenAI, and OpenRouter. The User must read and observe the terms, acceptable-use rules, and other published instruments of the supplier whose model the User selects. A breach of a supplier's current documents is a breach of this Policy, even if the same act is not separately listed in Section 1.4 or Section 5. The Administration may disable the account, refuse further requests, and withhold access to the affected models if it reasonably concludes that the User has violated a supplier document. The User may not circumvent a supplier restriction by changing models, by splitting a request, or by instructing the agent to ignore that restriction. Ignorance of a supplier's document does not excuse a violation. Official texts are those published by the supplier at the addresses stated below, as updated by that supplier from time to time.

7.2. Official documents of the suppliers

The following links point to the current public offers and related terms of the named suppliers. The User is responsible for opening the applicable document before sending a request to a model of that supplier. If a supplier replaces a URL, the document in force is the one that the supplier then designates as current. EntoCode reproduces these addresses for convenience and does not host or amend the suppliers' texts. Names and symbols of the suppliers are used for identification only and do not mean that EntoCode is affiliated with those companies. The Administration is not a party to the contract between the User and a supplier except where that supplier's terms so provide. Where several documents of one supplier apply, the User must comply with all of them that govern the relevant use.

7.3. Effect of supplier terms on use of the agent

Because requests are routed to supplier infrastructure as described in Section 6, the User's conduct toward a model is also conduct toward that supplier. Prohibited uses stated by a supplier, including restrictions on malware, abuse, and unlawful content, apply to prompts and tool results sent through EntoCode. The User automatically accepts this when selecting or using a model from the catalog. A supplier may refuse, log, or retain a request under its own documents; EntoCode cannot waive that supplier's rules. Conflict between this Policy and a mandatory term of a supplier is resolved by complying with both to the extent possible, and by not using the affected model if compliance is impossible. The Administration may update the list of links in clause 7.2 without a separate notice to each User. Questions about a supplier's document should be addressed to that supplier. Questions about how this section is applied inside EntoCode may be sent through the support channels listed on the EntoCode website.

8. Circumvention of restrictions

8.1. Debugging, reverse engineering, virtual machines, and decompilation

The Administration fully prohibits the use of debugging programs, reverse-engineering tools, virtual machines, or other decompilation instruments for the purpose of analyzing, bypassing, extracting, or defeating the protections, system instructions, billing, authentication, or access controls of EntoCode. This prohibition covers, without limitation, disassembly of the desktop application or website, interception of traffic in order to alter requests, extraction of hidden prompts or keys, and any laboratory or sandboxed setup whose object is to circumvent a restriction of the Service. Ordinary use of a virtual machine or debugger for the User's own lawful project, where that use is not directed at EntoCode's controls, is not by itself a violation of this clause. Use of such tools against the Service, its clients, or its control plane is a material breach. The Administration may treat preparation, documentation, or sharing of such methods as the same violation. Ignorance of a technical barrier does not authorize an attempt to remove it. The User must not instruct third parties to perform these acts on the User's behalf.

8.2. Prompts intended to bypass the prohibition of malware

The Administration fully prohibits the use of prompts, jailbreaks, role-play, encoding tricks, or any other wording designed to bypass the prohibition on creating viral and other malicious software stated in Section 1.4. A request that asks the agent to ignore safety rules, to pretend that malware is a hypothetical, educational, or fictional exercise, or to produce the same harmful result in fragments, is a violation of this clause. The fact that a model nevertheless returns output does not legalize the request or transfer responsibility to EntoCode. The User remains responsible for every such query as provided in Section 5.3. Attempts of this kind may be logged, refused, or reported as abuse. Repeated or sophisticated attempts aggravate the breach. If the User does not accept this prohibition, the User must not use the agent.

8.3. Unlawful use of the agent

The Administration fully prohibits use of the AI agent for unlawful acts that violate the Constitution of the Russian Federation, the legislation of the Russian Federation, or the current documents of the supplier companies named in Section 7. Unlawful use includes, without limitation, fraud, unauthorized access to systems, incitement prohibited by law, and any activity that those supplier terms forbid. Routing a request through EntoCode does not place that request outside the Constitution of the Russian Federation or outside a supplier's terms. The Administration may act on the content of prompts, tool results, and reports when assessing such use. A claim that the act was performed "only as text" or "only in a chat" does not exclude this clause. Consequences may include the measures in clause 4.4 and the measures in clause 8.6. The User must discontinue the request if compliance with those instruments is impossible.

8.4. Attempts to disable system instructions

The Administration fully prohibits attempts to disable, override, rewrite, or conceal the system instructions of EntoCode or of a connected model. This includes prompts that order the agent to ignore prior rules, to reveal hidden instructions, to enter a mode without restrictions, or to treat Administration policies as optional. It also includes technical steps intended to strip, replace, or intercept system prompts in the client or on the network. System instructions exist to apply this Policy, billing, and supplier requirements; they are not a puzzle for the User to remove. Success or failure of the attempt is irrelevant: the attempt itself is a breach. The Administration may immediately restrict the account upon detection. The User may not ask another person or another tool to disable those instructions in the User's place.

8.5. Attempts to obtain subscription abuse

The Administration fully prohibits attempts to obtain "abuse" of subscriptions, meaning any scheme to receive paid capacity, models, or credits without proper payment or beyond the limits of the User's plan. This includes, without limitation, sharing one paid account among many persons in order to evade plan limits, creating multiple accounts to multiply free or promotional credit, manipulating usage records, exploiting billing errors, or using stolen or unauthorized payment instruments. Circumvention of rate limits, credit counters, or plan gates by technical or prompt-based means is the same violation. Unused or disputed credits are not a license to extract additional usage. The Administration may cancel credits obtained by abuse and may refuse refunds where the account was used in breach. The User must use only the plan assigned to that User's verified account.

8.6. Immediate blocking

The Administration may immediately block the User for any one of the reasons listed in clauses 8.1 to 8.5, or for any other reason that the Administration considers sufficient to protect the Service, other Users, suppliers, or the law. Blocking may include disablement of sign-in, revocation of tokens, suspension of the agent, and withholding of paid features, without prior notice. The Administration is not obliged to disclose the full technical basis of the decision where disclosure would assist further circumvention. Unused credits are not a ground to keep an account active after a block under this section. A later review may confirm, modify, or lift the measure at the Administration's discretion. Re-registration from a new account in order to evade a block is itself a violation. Questions about a block may be sent through the support channels listed on the EntoCode website; submission of a question does not suspend the block.

9. Prohibition of reverse engineering, data scraping, and model extraction

9.1. Data scraping and automated collection

The Administration fully prohibits the use of EntoCode for the collection of data (scraping), automated parsing, mass harvesting of prompts or outputs, or any other systematic extraction of content from the Service or from the models available through the agent. This prohibition covers scripts, bots, queues, and any other means of sending a volume of requests whose purpose is to accumulate model answers rather than to perform ordinary development work. The User may not treat the agent as a data source for building a dataset. A high volume of requests that is not explained by bona fide use of the workspace may be treated as scraping. Circumvention of rate limits or credit gates in order to continue collection is a separate breach under Section 8. The Administration may immediately disable the account upon detection of such activity.

9.2. Reverse engineering and copying of application logic

The Administration fully prohibits reverse engineering, decompilation, or other analysis of EntoCode directed at copying the logic of the application, the agent, the routing of models, or the system of billing and access. This clause supplements Section 8.1 and applies equally to attempts to reconstruct hidden prompts, tool pipelines, or product architecture. The User may not reproduce, publish, or commercially exploit such logic. Ordinary study of the User's own project is not a violation. Analysis whose object is EntoCode itself, or the manner in which a supplier model is invoked through EntoCode, is a violation. Sharing the results of such analysis with third parties is the same breach. The Administration may treat preparation of such analysis as the completed violation.

9.3. Use of model outputs to train competing commercial systems

The Administration fully prohibits the use of answers obtained through EntoCode for the training, fine-tuning, distillation, or other improvement of competing commercial artificial-intelligence systems. This includes sending a large number of requests in order to teach another neural network on outputs associated with models such as those of Google LLC, OpenAI, Anthropic PBC, xAI, or other suppliers listed in the Service. Such use is also a breach of the supplier documents named in Section 7 and may cause those suppliers to disable the API keys used to operate EntoCode. The User may not sell, donate, or otherwise transfer harvested outputs for that purpose. A claim that the collected answers were used only internally does not exclude this clause if the purpose is to train a competing commercial system. The User remains responsible for every request as provided in clause 5.3. If the User does not accept this prohibition, the User must not use the agent.

9.4. Consequences

A breach of this section is a material breach of the Policy. The Administration may immediately block the User as provided in clause 8.6, refuse further service, and cancel unused credits. The Administration may also restrict or terminate access in order to protect supplier API keys and the continuity of the Service for other Users. The User shall indemnify the Administration against claims and losses arising from scraping, reverse engineering, or distillation carried out through the User's account, to the extent permitted by law. Nothing in this section authorizes an act that a supplier forbids. Questions about this prohibition may be sent through the support channels listed on the EntoCode website.

10. Responsibility for AI hallucinations and errors in code

10.1. Tendency of models to hallucinate

The models available through EntoCode are prone to "hallucinations": they may generate text, code, facts, citations, paths, and explanations that appear plausible but are incomplete, outdated, insecure, or untrue. The agent does not independently verify that a generated statement corresponds to the User's repository, to documentation, or to reality. A confident tone in a reply does not mean that the reply is correct. The Administration does not represent that any particular output is free of hallucination. The User must treat every generated result as unverified until the User has checked it. This clause applies to chat answers, diffs, commands, configuration, and any other material produced by a model. Ignorance of this tendency does not shift responsibility to EntoCode.

10.2. No warranty of generated code

The Administration does not guarantee the correctness, security, or operability of code generated by artificial intelligence through the Service. Generated code may contain critical vulnerabilities, logic errors, destructive commands, or defects that cause a server to fail or a database to leak. The Administration does not warrant fitness for a particular purpose, including production use. Display of code in the application, application of a patch by a tool, or the absence of an immediate error does not constitute approval of that code by the Administration. Section 5.3 allocates responsibility for queries; this section separately allocates responsibility for the result. The User accepts the risk that a generated result may be harmful or unusable. If the User requires a guaranteed result, the User must not rely on the agent as the sole author of production code.

10.3. Mandatory audit and testing before production

The User is obliged to conduct a mandatory audit (code review) and testing of any code obtained through the agent before running that code in production or on any system whose failure would cause material harm. The audit must include review of security, correctness, and fitness for the intended environment. Testing must be performed in a setting that does not expose production data or credentials unless the User has separately accepted that risk. The Administration does not perform this audit for the User and does not certify that a generated change is safe to deploy. Skipping review or testing is the User's decision and the User's risk. Automated application of a change by the agent does not replace the User's review. The User remains responsible for the decision to merge, execute, or publish the result.

10.4. Allocation of liability for the result

Responsibility for the consequences of using generated output, including downtime, data loss, security incidents, and claims of third parties, lies with the User and not with the Administration, except where mandatory law provides otherwise. The User shall indemnify the Administration against claims arising from the User's use of generated code, to the extent permitted by law. Nothing in this section limits the User's duties under Sections 1.4, 5, 7, and 8. The Administration is not liable for losses caused by a hallucination, an insecure snippet, or an erroneous command that the User chose to apply. This allocation applies even if the User asked the agent to "check" or "secure" the code. Questions about this section may be sent through the support channels listed on the EntoCode website.

11. Intellectual property in generated code

11.1. No claim by the Administration

The Administration of EntoCode does not claim intellectual-property rights in code, text, or other material created with the assistance of the agent solely by reason of that material having been generated through the Service. Operation of the website, the desktop application, and the proxy is not an assignment to EntoCode of the User's rights in the User's project. The Administration does not treat generated output as a work made for the Administration. This clause does not transfer to the User any rights in the EntoCode software, trademarks, or documentation. It does not authorize scraping or distillation prohibited by Section 9. Names and symbols of suppliers are used for identification only and do not mean affiliation.

11.2. Rights of the User

Subject to clause 11.3, all rights in code created with the assistance of the agent that can belong to the User belong to the User. The User may use, modify, and dispose of such code within the User's project as the User sees fit, without a separate license from EntoCode. The Administration does not require the User to attribute generated fragments to EntoCode. This allocation is intended to allow developers and companies to use the agent without EntoCode asserting ownership of the resulting code. The User remains responsible for ensuring that the User's own contributors, employers, or clients agree to this use. Nothing in this section obliges the Administration to register, defend, or enforce the User's rights. If a dispute arises between the User and a third party about authorship of generated code, that dispute is not a dispute about a claim by EntoCode.

11.3. Limits set by model suppliers

The rights described in this section exist only to the extent that the current documents of the model suppliers named in Section 7, including Google LLC, OpenAI, Anthropic PBC, xAI, and OpenRouter, permit the User to own or use the relevant output. Some suppliers restrict the use of outputs, require reservation of their rights, or forbid certain commercial uses. EntoCode cannot grant a broader right than the applicable supplier grants. The User must read those documents before relying on generated code as exclusively the User's property. Conflict between this section and a mandatory supplier term is resolved by the supplier term for that output. The Administration does not warrant that any supplier will treat a given output as free of third-party claims. If the User does not accept those limits, the User must not use the affected model.

12. Contact

Privacy requests can be sent through the support channels listed on the EntoCode website.