Argosvix

Terms of Service

Last updated: 2026-07-02 Effective: 2026-07-02

These Terms of Service ("Terms") govern your use of "Argosvix" (the "Service") operated by Yuto Makihara ("we", "us"). By using the Service you ("you") agree to these Terms.

Article 1 (Purpose and Service Description)

  1. These Terms set out the conditions for use of "Argosvix" (the "Service") provided by us. They apply to all relationships between users and us regarding use of the Service.

  2. The Service is infrastructure that collects, stores, and aggregates records of API calls (the "Call Records") sent via our SDK from AI agents and LLM applications operated by the user, and provides visualization and statistics to the user. In principle, Call Records cover only metadata (provider name, model name, token counts, cost, latency, error information, user-defined tags, etc.). Users on the Pro plan or higher may enable, by explicit consent, the additional feature defined in Article 4-2 (the "Plaintext Storage Feature") and, separately, may enable by explicit consent the "AI Helper Feature" defined in Article 4-3 (automatic safety classification, evaluation criteria proposal, and plaintext PII secondary audit). Delivery transports are the SDK (npm package), the Python SDK, the Public REST API, the MCP server, the management console (web dashboard), and the chat feature within the management console (Article 4-4); details follow the documentation within the Service.

  3. The Service does not provide real-time anomaly detection, individualized advice on the user's AI model selection, prompt design, or production decisions, or any direct service-level guarantee to the user's end users.

Article 2 (Registration)

  1. Use of the Service requires account registration ("Registration") by the method we prescribe. To register, you must agree to these Terms and the separately established "Privacy Policy".

  2. The Service is intended for persons aged 18 or older. A minor under 18 may use it only with the consent of a guardian, and the guardian is deemed to have agreed to these Terms regarding the minor's use.

  3. We may decline Registration if we determine the applicant falls under any of the following, with no obligation to disclose the reason:

    • The applicant has previously violated these Terms
    • The application contains false statements
    • We otherwise determine Registration to be inappropriate

  4. You are solely responsible for managing your account and API keys, and must not let a third party use them or lend, transfer, rename, or sell them. We are not liable for damages arising from leakage or unauthorized use of API keys.

Article 2-2 (OAuth Authentication)

  1. As an alternative to registering an email address and password, you may register using OAuth 2.0 authentication provided by GitHub Inc. or Google LLC.

  2. Through OAuth authentication we obtain your provider-verified email address, the identifier within the provider (the "subject"), and display name. Handling of this information follows the separately established "Privacy Policy".

  3. If you already hold an account by password authentication with the same email address, new registration via OAuth will result in an error. In that case we guide you to log in by password authentication and then add OAuth linking from the settings screen (this feature is provided progressively).

Article 3 (Fees)

  1. The Service offers a free plan (Free) and paid plans (Pro / Team). Details of fees, payment methods, scope of provision, and other conditions follow the Pricing page within the Service. Where such description differs from these Terms, these Terms prevail.

  2. Paid plans are charged automatically according to the billing cycle of the plan applied for (currently monthly only; if we offer an annual cycle, it will be clearly indicated at the time of application) using the payment method (credit card) registered at the time of application. Payment is processed by Stripe Inc. (USA).

  3. You may cancel at any time from our Stripe Customer Portal. Cancellation takes effect from the next renewal date, and you may continue to use the Service until that renewal date.

  4. Paid plans may include a free trial period on a per-plan basis. The availability and length of any trial period follow the display on the Pricing page at the time of application (as to this point, this paragraph constitutes an exception to the order of precedence in Paragraph 1). If you cancel during the trial, no charge occurs. After the trial ends, unless you cancel, the registered payment method is charged automatically. At the start of the trial we clearly indicate the trial end date and the automatic billing start date and notify you by email.

  5. Because the Service is the provision of digital services, fees once paid are not subject to the cooling-off provisions under Article 15-3 of the Act on Specified Commercial Transactions (cancellation in mail-order sales). Except in cases of serious cause attributable to us, fees are non-refundable. Treatment where the Service becomes unavailable due to outage or maintenance follows Article 6.

  6. We may change the fees, payment methods, scope of provision, and other conditions of the Service. For material changes, we will notify the content and date of the change within the Service in advance. Where the change is disadvantageous to users, we will notify it at least 30 days before the effective date.

Article 4 (Data Handling)

  1. You consent to send us, through the Service, the Call Records (including provider name, model name, token counts, cost, latency, error information, and user-defined tags) arising from AI agents and LLM applications you operate.

  2. You warrant the following regarding the content of the Call Records you send us:

    • You have the necessary rights to collect, send, and store such Call Records
    • Such Call Records do not contain personal information of your end users (name, email address, address, phone number, My Number, payment information, etc.; excluding, however, pseudonymized user identifier tags attached by the SDK's standard functionality)
    • If you enable the Plaintext Storage Feature defined in Article 4-2, the plaintext data (prompt body, AI completion body, tool call arguments and results) does not contain end users' personal information, confidential information, trade secrets, or content that infringes third-party intellectual property rights
    • Sending such Call Records to us does not violate any contract or rules with your organization or counterparties

  3. If you violate the preceding paragraph 2 and receive a claim from a third party arising from the content of the Call Records sent to us, you shall resolve such claim at your own expense and responsibility and hold us harmless.

  4. The purpose, scope, retention period, and presence of third-party provision for our handling of Call Records follow the separately established "Privacy Policy".

  5. We do not use the Call Records you send us (metadata and, when the Plaintext Storage Feature is enabled, the plaintext) for our AI model training, business-to-business data sales, or provision for purposes other than those of the user concerned. However, this does not apply to disclosures required by law, where necessary to protect the life, body, or property of a person, or where you have given prior consent.

Article 4-2 (Plaintext Storage Feature)

  1. You may, only while you hold a subscription to the Pro plan or higher (the "Eligible Plans"), enable the "Plaintext Storage Feature" by explicit opt-in from the settings screen of the dashboard. The default setting is disabled (plaintext is not stored).

  2. When you enable the Plaintext Storage Feature, we store the prompt body, AI completion body, and tool call arguments and results (the "Plaintext Data") that you transmit through the SDK, encrypted using AES-256-GCM, for the retention period you select (7, 30, or 90 days, as available on your subscribed plan).

  3. To enable the Plaintext Storage Feature, you must understand and agree to the purpose of use, retention period, encryption method, and right to delete shown in the consent dialog (the "Consent Dialog") displayed in the settings screen. We retain the date and time of consent, the source IP address, the User-Agent, and the version of the Consent Dialog, and you may review these from the settings screen.

  4. The purpose of use of the Plaintext Data is limited to the user's own review of AI output quality, annotation, prompt improvement, and evaluation execution. We do not use the Plaintext Data for our AI model training, aggregate statistics, or provision to third parties (excluding infrastructure providers such as Cloudflare engaged as subcontractors).

  5. You may stop the Plaintext Storage for subsequent calls by turning the toggle off in the settings screen. After switching off, the already stored Plaintext Data remains until the retention period in Article 4-2 Paragraph 2 expires. You may delete the stored Plaintext Data via the "Bulk delete plaintext records" button in the settings screen or the delete button on each individual record.

  6. We record every decryption access to the Plaintext Data (including access by our operators) in an audit log, and you may review the access log for your Plaintext Data from the settings screen. When our operators need to decrypt the Plaintext Data for purposes such as customer support, we obtain your individual prior consent and do not decrypt without consent.

  7. As a standard SDK configuration, we enable by default a filter that masks email addresses, credit card numbers, phone numbers, My Number, and IP addresses before transmission. You may disable this masking at your own responsibility, but you bear the risk of personal information leakage if disabled.

  8. If you cancel the Eligible Plans, all Plaintext Data associated with your account is physically deleted within 30 days from the cancellation date. If you re-subscribe to the Eligible Plans within 30 days of cancellation, the deletion is not executed.

Article 4-3 (AI Helper Feature)

  1. The "AI Helper Feature" collectively refers to the following features we provide to users subscribed to the Pro plan or higher (the "Eligible Plan Users"):

    • Automatic safety classification (when the user has enabled the Plaintext Storage Feature, sending a portion of the plaintext to the OpenAI Moderation API to classify whether it contains inappropriate content)
    • Evaluation criteria proposal (when the user requests it via the MCP server or the Public REST API, sending excerpts of plaintext call records selected by the user together with a "use case" text to the GPT-4o-mini model to generate candidate evaluation criteria suitable for that use case)
    • Plaintext PII secondary audit (when the user has enabled the Plaintext Storage Feature, sending a portion of the plaintext to the GPT-4o-mini model to detect inclusion of personal information and recording the detection results for the user's review)
    • Automatic quality evaluation (when the user requests an evaluation run, or when we periodically run automatic scoring to detect changes in the quality of AI outputs, sending a portion of the plaintext to the GPT-4o-mini model to generate scoring results and scoring rationales against evaluation criteria and recording them for the user's review)

  2. The AI Helper Feature is enabled only when an Eligible Plan User provides explicit consent via a dedicated consent dialog ("AI Helper Consent") from the settings screen of the dashboard. The default setting is disabled, and consent is obtained independently of the consent to the Plaintext Storage Feature.

  3. To enable the AI Helper Feature, the Plaintext Storage Feature (Article 4-2) must also be enabled. If the Plaintext Storage Feature is not enabled, the AI Helper Feature does not activate.

  4. To implement the AI Helper Feature, we transmit a portion of the Eligible Plan User's plaintext data (up to 1,500 characters each of prompt body and AI completion body per call) to OpenAI, Inc. (USA, our subprocessor for processing personal data). Transmission destinations are limited to the following:

    • For automatic safety classification: the OpenAI Moderation API (a model specialized for inappropriate-content detection such as omni-moderation-latest)
    • For evaluation criteria proposal: the OpenAI Chat Completions API (the gpt-4o-mini model)
    • For plaintext PII secondary audit: the OpenAI Chat Completions API (the gpt-4o-mini model)
    • For automatic quality evaluation: the OpenAI Chat Completions API (the gpt-4o-mini model)

    We may change the destination models set out in this paragraph to successor equivalent models, to the extent that conditions equal to or stricter than those in Paragraph 5 of this Article (exclusion from AI model training and limits on log retention periods) are satisfied. Any such change will be made known through a revision of these Terms or the Privacy Policy.

  5. We require our subprocessor OpenAI, Inc. to comply with the following under industry-standard API terms (including OpenAI Enterprise privacy commitments):

    • Not use received data to train OpenAI, Inc.'s AI models
    • Delete received data from logs within at most 30 days, and discard within that period absent legal or other justification
    • Process such data in accordance with our instructions and not use it for other purposes

  6. If you revoke AI Helper Consent, we will not transmit your plaintext data to OpenAI, Inc. for any new processing request that starts at or after the time of revocation. Revocation takes effect immediately via the toggle on the dashboard settings screen, and we promptly stop transmission for new requests after revocation. For requests already in progress at the time of revocation ("in-flight requests"), processing may continue until that request completes; however, to the extent reasonably implementable, we re-verify the consent state immediately before completing such an in-flight request and stop processing if the revocation has taken effect. Handling of data already transmitted before revocation follows paragraph 5 of this Article.

  7. The AI Helper Feature is automatically disabled at the same time as cancellation of the Eligible Plans. AI Helper Consent is re-obtained upon re-subscription after cancellation.

  8. API usage fees of OpenAI, Inc. arising from use of the AI Helper Feature are borne by us. No additional charge is made to the user beyond the plan fee (provided that we may impose reasonable limits on the frequency of use of the feature via the Service's internal rate limits).

  9. You acknowledge that outputs of the AI Helper Feature (safety classification labels, candidate evaluation criteria, plaintext PII secondary audit results, and automatic quality evaluation scoring results) are based on the statistical inference of a large language model, and we do not warrant the accuracy, completeness, or fitness for a particular purpose of the outputs. Generated outputs are positioned as reference information to assist your own judgment; the Service does not automatically take other actions (billing, notification, account suspension, etc.) based on those outputs.

Article 4-4 (Chat Feature)

  1. The "Chat Feature" refers to the interactive feature within the management console by which the user enters questions in natural language and we generate responses by referring to statistics of already-aggregated Call Records and similar information.

  2. To generate Chat Feature responses, we transmit the question text entered by the user, the context of the conversation, and aggregated metadata to the extent necessary to generate the response (statistics of Call Records, summaries of configuration information, etc.) to the Chat Completions API of OpenAI, Inc. (USA) (the gpt-5.4-mini model; the successor-model change provision in the latter part of Article 4-3, Paragraph 4 also applies to this Article). We do not transmit the body of Plaintext Data (Article 4-2) via the Chat Feature.

  3. As to use of the Chat Feature, the act of submitting a question in the management console is deemed to constitute consent to the transmission set out in this Article. The treatment of the transmitted data at OpenAI, Inc. (exclusion from AI model training and limits on log retention periods) is subject to the same conditions as Article 4-3, Paragraph 5.

  4. Chat conversation history is stored within the Service while you hold an account. You may archive history on a per-conversation basis from the management console, and upon account withdrawal, conversation history is deleted as data associated with your account.

  5. Chat Feature responses are reference information based on the statistical inference of a large language model; as with Article 4-3, Paragraph 9, we do not warrant their accuracy, and the Service does not automatically perform billing, configuration changes, or other operations based on a response (except where you explicitly approve an operation contained in a response).

  6. You may register your own OpenAI API key in the dashboard settings and have chat responses generated with that key ("bring-your-own-key"). When registered: (1) OpenAI Inc. usage fees for response generation are billed to you under your contract with OpenAI Inc., and our monthly chat usage cap does not apply; (2) the key is stored encrypted with AES-256-GCM and is never used for any purpose other than response generation; (3) you may delete the key at any time from the dashboard, after which the default service-funded behaviour resumes; (4) if the key becomes invalid, chat returns an error and never silently falls back to the service-funded behaviour.

Article 5 (Prohibited Acts)

You must not engage in any of the following when using the Service:

  1. Acts that violate laws or public order and morals
  2. Acts related to criminal activity
  3. Acts that destroy or interfere with the functioning of our servers or network
  4. Reverse engineering, decompiling, or disassembling the Service
  5. Disclosing others' information obtained through the Service to third parties without the consent of those others
  6. Excessive API calls, circumventing rate limits, or imposing improper load
  7. Reselling, subleasing, or using the Service to build a competing service without our prior written consent
  8. Enabling the Plaintext Storage Feature and, in violation of Article 4 Paragraph 2, transmitting to us content that infringes end users' personal information, confidential information, trade secrets, or third-party intellectual property rights
  9. Providing consent without genuine intent (such as checking the boxes without reviewing the content) to the consent items shown in the Consent Dialog of the Plaintext Storage Feature, and subsequently bringing a dispute with us regarding the handling of such data
  10. Other acts we deem inappropriate

Article 6 (Suspension, Change, and Termination of the Service)

  1. We may suspend all or part of the Service without prior notice to users in any of the following cases:

    • Maintenance or updates of the Service
    • Force majeure such as earthquake, lightning, fire, power outage, or natural disaster makes provision difficult
    • Computers or communication lines stop due to an accident
    • We otherwise determine that provision of the Service is difficult

  2. We may permanently terminate the Service. In that case we will notify users at least 90 days before the planned termination date and provide a reasonable opportunity for users to export their data.

  3. We are not liable for damages incurred by users or third parties due to suspension, change, or termination under this Article, except in cases of our intent or gross negligence. The scope of our liability and the cap on damages in such case follow Article 7. For paid plan subscribers, we will refund the unused portion calculated on a pro-rata daily basis for the suspension or termination period.

Article 7 (Disclaimer)

  1. We do not warrant, expressly or impliedly, that the Service is free from defects in fact or law (including safety, reliability, accuracy, completeness, validity, fitness for a particular purpose, absence of harmful components such as viruses, and absence of infringement of others' rights).

  2. Except in cases of our intent or gross negligence, we are not liable, among damages incurred by users arising from the Service, for damages arising from special circumstances (including where we foresaw or could have foreseen such circumstances), lost profits, business interruption, damages arising from loss or corruption of data, and indirect damages. The cap on damages where we bear liability is as set out in the next paragraph.

  3. Where we bear liability for damages to a user arising from the Service, except in cases of our intent or gross negligence, the amount of damages is capped at the total fees the user paid us for the Service in the most recent 12 months (or ¥10,000 for the free plan).

Article 8 (Changes to These Terms)

  1. Pursuant to Article 548-4 of the Civil Code of Japan, we may change these Terms without the individual consent of users.

  2. When we change these Terms, we will make known the effective time and the content of the changed Terms by posting within the Service or other appropriate means.

  3. Where the change does not conform to the general interests of users, we will notify the content at least 30 days before the effective date, and users may avoid the effect of the change by canceling before the effective date.

Article 9 (Handling of Personal Information)

Handling of personal information we obtain in connection with use of the Service follows the separately established "Privacy Policy".

Article 10 (Governing Law and Jurisdiction)

  1. These Terms are governed by and construed in accordance with the laws of Japan.

  2. Any dispute relating to the Service shall be subject to the exclusive jurisdiction of the court having jurisdiction over our location as the court of first instance.