Special Provisions Regarding MCP Agent Transactions
This document is made in Japanese and translated into English for reference purposes only. In the event of any discrepancy between the Japanese and English versions, the Japanese version shall prevail.
These special provisions (hereinafter referred to as "these special provisions") define the terms of use for the function (hereinafter referred to as "this function") that allows users to access the MCP server managed by Woodstock K.K. (hereinafter referred to as "the Company") using an AI agent selected by the user to obtain transaction information and conduct transactions, etc., within the financial product brokerage service provided by the Company via the Internet. By commencing use of this function, the user is deemed to have agreed to all the contents of these special provisions.
Article 1 (Purpose and Relationship with Original Terms)
These special provisions, as special provisions to the "Terms of Use" (hereinafter referred to as "Original Terms") established by the Company, aim to define the rights and obligations between the user using this function and the Company.
Regarding the use of this function, in addition to the matters stipulated in these special provisions, the provisions of the Original Terms shall also apply in conjunction.
In the event of any conflict or inconsistency between the provisions of these special provisions and the provisions of the Original Terms, the provisions of these special provisions shall prevail.
Article 2 (Definitions)
1. The following terms used in these special provisions shall have the meanings set forth in the following items. (1) "MCP (Model Context Protocol)" refers to an open protocol standard for seamless data exchange and tool execution between clients such as AI agents and servers such as data sources.
(2) "MCP Server" refers to a server system built and managed by our company to provide this function, compliant with the MCP standard.
(3) "AI Agent" refers to the entire MCP client software, program, or system that connects to the MCP server and functions autonomously or interactively, including external large-scale language models (LLMs), which are selected and used by the user at their own discretion and expense. Depending on the context, it may also refer to the provider of the external LLM processing server.
(4) "OIDC Integration" refers to the connection integration function that uses the OpenID Connect protocol to verify the user's identity (authentication) and grant access rights (authorization) to specific AI agents.
(5) "Authorization Token" refers to electronic key information such as access tokens and refresh tokens issued based on OIDC integration, which are necessary for AI agents to access the MCP server. 2. The definitions of terms used in these Special Provisions, other than those defined in these Special Provisions including the preceding paragraph, shall be as set forth in the original Terms and Conditions.
Article 3 (Connection Authorization via OIDC Integration and User Management Responsibilities)
When using this function, the user shall perform identity verification using OIDC integration and authorization of data access and transactions for a specific AI agent (hereinafter referred to as "Authorization Agreement") on the screen designated by the Company.
By the user clicking "Agree" or a similar button on the Authorization Agreement screen mentioned in the preceding paragraph, the Company will treat the user as having definitively granted the AI agent the authority to acquire transaction information and execute orders based on their account.
The user shall strictly manage the authorization token issued in connection with OIDC integration and stored and managed on the AI agent side under their own responsibility, and shall not disclose, lose, or allow any third party (including AI agent providers and other third parties not anticipated by the user) to misuse it.
Access to the MCP server and transaction instructions using the authorization token shall be considered valid actions performed by the user themselves, regardless of the reason. We shall not be liable for any damages or losses incurred by the user due to insufficient management of the authorization token, errors in use, or unauthorized use by a third party.
If a user has any doubts about the security, operation status, or token management status of the AI agent with which they have linked, the user is obligated to immediately terminate the OIDC linkage (expiration of the authorization token) through the application, website provided by us, or other methods determined by us. Furthermore, the results of any transactions that occurred before the termination of the linkage shall all be attributed to the user, and we shall not be liable for any consequences thereof.
Article 4 (Attribution of Investment Decisions and Denial of Discretionary Investment Management)
All order commands, trading instructions, and data acquisition requests transmitted from the AI agent to our MCP server, regardless of the reason, shall be considered as the user's own definitive investment decisions and instructions, generated based on the parameters, instructions (prompts, etc.) set by the user, and the inferences of the AI agent selected by the user.
Even if the user selects a specification in which the AI agent autonomously completes the decision-making and order-ordering process, the user acknowledges in advance that all legal and economic effects related to such transactions shall directly belong to the user. Our company and affiliated financial instruments business operators merely provide the infrastructure to mechanically process orders and other instructions transmitted from AI agents via the MCP server and relay them to the backend system. We do not enter into any contract with users to act as agents for investment decisions, nor do we undertake investment management, investment advisory/agency services, or any other investment decision-making activities.
Article 5 (Acknowledgment of AI's Technical Characteristics and Disclaimer)
Users fully understand that AI agents (including external LLMs, etc.) are subject to uncertainties and technical limitations, such as hallucination (generation of false or misleading information), misinterpretation of instructions, program bugs, and prompt injection (unexpected behavior due to the loading of malicious instructions from third parties). Users shall use this function after accepting the risks associated with these limitations.
Our company shall not be liable for any damages, losses, or disadvantages incurred by users due to the following reasons (including, but not limited to, unexpected order execution, unintended failure of trades, losses due to price fluctuations, errors in quantity, amount, etc.).
(1) Incorrect orders resulting from AI agent malfunction, misinterpretation, or hallucination.
(2) The AI agent sending fraudulent order commands due to prompt injection or other cyberattacks by a third party.
(3) Actions resulting from deficiencies in user-set prompts, system instructions, or usage restrictions.
(4) Delays, failures, or inability to process orders due to internet congestion, connection problems, protocol malfunctions, or API rate limitations.
(5) Restrictions, suspensions, or delays in trading due to system reasons at the affiliated financial instruments business operator.
Article 6 (Nature of Data Disclosure and Privacy Exemption)
We may consider that the user has granted the AI agent the right to acquire personal information upon authorization and consent. Based on OIDC linkage, data transmitted from the MCP server to the AI agent (including the user's account balance, holdings, trading history, personal information, etc.) will be treated as disclosed and provided to the user based on the user's own authorization and consent.
Once data is transmitted from the MCP server to the AI agent, such data is outside the scope of our security management obligations and control.
Regarding data after transmission, we have no obligation to monitor for data leakage, misuse, unauthorized use for further training of AI models, or other violations of the privacy policy by AI agent providers, LLM providers, or other third parties, and we shall not be liable for any damages incurred by users or third parties as a result. Users are responsible for reviewing and accepting the data usage policies of each AI agent and LLM provider before using the service.
Article 7 (Restrictions Based on AI Governance)
To prevent fatal asset damage to users and disruption of financial markets caused by malfunctions of this function, AI agent malfunctions, unauthorized access, or abnormal orders, we may, at our discretion, set, modify, or apply the following trading restrictions (hereinafter referred to as "guardrails") on the MCP server side. [MH: We understand that such guardrails are in place based on the AI discussion paper published by the FSA.] If such measures are not in place, a human-in-the-loop system for order confirmation would be required. However, since we understand this is an automated order placement system, we naturally assume that such a system is already in place. We believe there can be various variations in the details of guardrail measures, so please inform us if you have implemented measures different from the typical guardrail measures listed below.
(1) Limitation on the maximum order amount or quantity per order
(2) Limitation on the total order amount, quantity, or number of orders per day
(3) Restriction on tradable securities
(4) Restriction on trading hours
(5) Other restrictions deemed necessary by our company to ensure secure trading
Our company shall not be liable for any loss of opportunity, exchange rate fluctuations, or any other damages and expenses incurred by the user as a result of the AI agent's orders failing, being rejected, or being delayed due to the activation of these guardrails.
Article 8 (Suspension, Modification, and Termination of the Function)
The Company may temporarily suspend, restrict, modify, or terminate (hereinafter referred to as "Suspension, etc.") all or part of the Function without prior notice or warning to the user in the event of any of the following circumstances:
(1) When performing regular or emergency maintenance of the MCP server or related systems
(2) When a serious vulnerability or system bug related to the AI agent or external LLM is discovered
(3) When there is administrative guidance, order, or request from regulatory authorities, or when laws or guidelines are enacted or amended
(4) When the Company otherwise determines that it is difficult or inappropriate to continue providing the Function
The Company shall not be liable for any damages, losses, or disadvantages incurred by the user or a third party as a result of the suspension, etc. of the Function pursuant to this Article.
Article 9 (Miscellaneous Provisions)
Regarding the terms of use of the Function that are not stipulated in these Special Provisions, the provisions of the original Terms of Service shall apply.
End
【Established: June 5, 2026】