Last updated: July 30, 2026

The SingChat open-source programme has officially begun.

This is not merely an open project for a standalone AI customer-support product. It is also an important part of the SingLinkVPN open-source and transparency programme.

SingChat is the live customer-support and AI assistance system currently used by SingLinkVPN. When SingLinkVPN users encounter questions about accounts, subscriptions, free data, nodes, connections, system compatibility, client updates, or technical troubleshooting, they can use SingChat to receive an initial AI answer, retrieve official knowledge, request human takeover, and obtain follow-up support.

SingLinkVPN's open-source and transparency work should not cover only VPN protocols, client architecture, performance tests, security models, and the no-logs policy.

The customer-support system that interacts directly with users also raises questions such as:

  • which operations AI may perform;

  • which data customer-support staff may view;

  • how workspaces are isolated;

  • how third-party AI tools connect;

  • how data voluntarily submitted by users is handled;

  • which capabilities are already public;

  • which parts remain under security and authorization review.

Including SingChat in the SingLinkVPN open-source and transparency programme therefore means that the scope of disclosure is expanding beyond underlying VPN technology to the customer-support and after-sales systems that users actually interact with.

As the first phase, SingChat has published the singchat-mcp repository on GitHub. Its public MCP manifest, tool descriptions, integration examples, and multilingual technical documentation are released under the MIT License. Any AI assistant that supports MCP can use SingChat MCP to create a workspace, obtain website embed code, manage support conversations, add AI knowledge, and view analytics. GitHub

SingChat open-source and product status summary

Item

Current status

Product name

SingChat

Product positioning

Cross-platform AI customer support and live chat system

Relationship to SingLinkVPN

Customer-support and AI assistance system currently used by SingLinkVPN

Programme positioning

Part of the SingLinkVPN open-source and transparency programme

SingChat MCP

Public GitHub repository available

MCP license

MIT License

Published material

MCP manifest, tool descriptions, integration examples, and multilingual documentation

MCP tool count

17

Supported AI integrations

Claude, Cursor, ChatGPT, Grok, Gemini CLI, and others

Website integration

Add the chat widget with one script

Main support model

AI answers, knowledge-base retrieval, and human takeover

Community Edition

Officially identified as AGPL-3.0

Enterprise Edition

Commercial license

Deployment

Official SaaS or self-hosting

Supported environments

Web, iOS, Android, Flutter, and other products

Future direction

Phased disclosure based on security, privacy, licensing, and product stability

The official SingChat website positions the product as an AI customer-support platform that can be embedded in Web, iOS, Android, and Flutter products. It provides a RAG knowledge base, human takeover, a shared inbox, multi-tenant isolation, APIs, webhooks, and a choice between the official SaaS and self-hosting. SingChat

1. SingChat is the customer-support system currently used by SingLinkVPN

SingChat was not created as a demonstration project for an open-source announcement. It is already part of SingLinkVPN's real user-support workflow.

SingLinkVPN operates across macOS, Windows, iOS, Android, Linux, TV, and the web. User questions may involve:

  • client downloads and installation;

  • version updates;

  • free data and plan rules;

  • account and order enquiries;

  • VPN node selection;

  • connection failures;

  • network-operator restrictions;

  • system and device compatibility;

  • security and privacy;

  • technical diagnostics voluntarily submitted by a user;

  • instructions and feature explanations.

In these situations, SingChat provides:

  • AI answers to frequently asked questions;

  • retrieval of official documents and announcements;

  • multilingual support;

  • human-agent takeover;

  • conversation-state management;

  • support entry points for websites and clients;

  • issue classification and follow-up handling;

  • knowledge-base updates and FAQ management.

SingChat's AI can first answer common questions from an organization's knowledge base. When it cannot handle a case or the case requires human judgement, it can transfer the conversation to a human support agent. SingChat

SingLinkVPN's real operating environment gives SingChat a practical validation environment involving multiple platforms, multiple languages, and complex technical questions.

However, SingChat does not serve only SingLinkVPN.

It is also an independent AI customer-support product that can be made available to other websites, apps, SaaS platforms, developer tools, and enterprise services. SingLinkVPN is an important real-world SingChat user and one of the use cases driving disclosure of SingChat's integration methods, tool permissions, and technical information.

2. Why is a customer-support system also part of VPN transparency?

When people discuss VPN transparency, they usually think first about whether:

  • the VPN logs browsing activity;

  • it logs DNS queries;

  • it stores source IP addresses;

  • the VPN protocol is public;

  • the client can be reviewed;

  • nodes handle network traffic appropriately;

  • performance data can be reproduced;

  • security reports contain sufficient evidence.

But data interaction between a user and a VPN service does not occur only inside the VPN tunnel.

When users encounter a problem and voluntarily contact support, they may submit:

  • a description of the problem;

  • a Mac, Windows, or mobile-device model;

  • an operating-system version;

  • a SingLinkVPN client version;

  • an error message;

  • a client screenshot;

  • order-enquiry information;

  • node and connection details;

  • diagnostic information voluntarily supplied by the user;

  • conversations with customer-support or community-support personnel.

Although this data is not browsing-log data automatically collected by the VPN, it still requires clear purpose limitations, access controls, and deletion rules.

A customer-support system must therefore answer:

  • Which support operations can the AI assistant perform?

  • Can AI tools read other workspaces?

  • Which information can customer-support personnel view?

  • How are API Keys bound to workspaces?

  • How are third-party AI tools authorized?

  • How are support conversations stored and deleted?

  • Are diagnostics submitted voluntarily by users?

  • Which system components are already public?

  • Which content is still not public?

Publishing SingChat MCP is SingLinkVPN's first step in answering these questions.

The current public documentation identifies the MCP endpoint, integration methods, 17 tools, and workspace-permission design. It also states that each tool operates only within the workspace bound to its API Key and does not accept an arbitrary tenant parameter. GitHub

This gives external developers and the community a clearer view of what an AI assistant can do and how the corresponding workspace boundary is defined.

3. The SingLinkVPN open-source and transparency programme covers more than VPN protocols

The SingLinkVPN open-source and transparency programme is progressively covering different product layers:

SingLinkVPN open-source and transparency programme
├─ Public VPN architecture and technical documentation
├─ Security and privacy model
├─ No-logs policy and third-party verification
├─ Performance-test methods and data formats
├─ Future disclosure of the SingLink protocol and clients
├─ SingChat support and MCP integration
├─ Open-source verification tools
└─ Vulnerability disclosure and community oversight

SingChat is not a separate project unrelated to the SingLinkVPN open-source programme.

It shows that SingLinkVPN is extending disclosure beyond VPN transport, clients, security, and performance systems to:

  • user support;

  • AI customer support;

  • knowledge bases;

  • MCP integration;

  • after-sales conversations;

  • workspace permissions;

  • third-party AI tool integration.

For SingLinkVPN, transparency at only the VPN transport layer, while leaving the support system completely closed, would still not constitute complete product transparency.

When users voluntarily give account, order, device, or diagnostic information to customer support, the purpose, permissions, and technical boundaries of the relevant system also deserve a clear explanation.

4. What is SingChat?

SingChat is an AI customer-support platform for websites, apps, and enterprise systems.

It places traditional live support and AI support in the same workspace:

  • a visitor starts a conversation through a website or app;

  • AI answers from the organization's knowledge base;

  • a human agent can take over the same conversation;

  • a team can assign, reply to, and resolve tickets;

  • an administrator can review conversations and AI usage;

  • a developer can manage the system through APIs, webhooks, and MCP.

SingChat is not merely a chat button placed in the lower-right corner of a website.

Capabilities listed by the official website include:

  • AI customer support and a RAG knowledge base;

  • a live-chat widget;

  • human takeover;

  • a shared team inbox;

  • internal notes;

  • conversation assignment and resolution;

  • business hours and offline replies;

  • multi-tenant workspaces;

  • roles and permissions;

  • REST APIs;

  • scoped API Keys;

  • webhooks;

  • an OpenAPI specification;

  • WebSocket real-time communication;

  • official SaaS;

  • self-hosted Community Edition. SingChat

5. Why was SingChat developed?

Traditional support systems are usually built around an administration panel operated by human support agents.

An organization first has to register an account, create a workspace, configure a website, copy code, invite agents, and then separately configure AI, a knowledge base, and automated replies.

This model creates several common problems.

1. AI and human customer support are separated

Some support platforms treat AI as an optional plug-in.

AI answers, human tickets, knowledge, and visitor data may be spread across different locations, forcing a human agent to reconstruct the issue when taking over.

SingChat aims to let AI and humans use the same conversation. AI handles questions that can be standardized; when a case needs human attention, an agent sees the existing context and continues the reply directly.

2. Website, app, and other entry points are fragmented

A website may use one support system, a mobile app another, and a desktop client yet another.

Through its website widget, WebView, APIs, and cross-platform integration methods, SingChat lets different products centralize support in one workspace. The official site currently positions SingChat as a cross-platform support system for Web, iOS, Android, and Flutter products. SingChat

3. Traditional administration panels are not designed for AI assistants

Without MCP, users usually have to operate the support administration panel one page at a time.

With MCP, an AI assistant can call explicitly authorized customer-support tools directly—for example, creating a workspace, obtaining embed code, adding knowledge, or managing conversations.

4. Platform lock-in

When support data, knowledge, and integration methods are completely tied to a single SaaS product, migration costs increase over time.

SingChat provides APIs, webhooks, OpenAPI, and a self-hosting path so that developers and organizations retain more technical choices. SingChat

5. Data and deployment-control requirements

Some organizations need to deploy a customer-support platform on their own servers or in a designated region and control their own database, attachments, AI models, and backups.

Official SingChat documentation positions Community Edition as a free, self-hosted edition under AGPL-3.0. SingChat

6. What exactly has SingChat made public?

“SingChat open source” must be understood accurately for each component.

The current scope can be divided into three layers.

Layer one: the public SingChat MCP repository

The public SingLinkLabs/singchat-mcp repository contains:

  • the public MCP manifest;

  • MCP tool descriptions;

  • Claude Code configuration;

  • Cursor configuration;

  • ChatGPT integration instructions;

  • Grok integration instructions;

  • Gemini CLI configuration;

  • website-embedding examples;

  • multilingual README files;

  • the MIT License. GitHub

The repository currently provides documentation in Traditional Chinese, Simplified Chinese, English, Japanese, Korean, French, Spanish, Portuguese, Arabic, Russian, Thai, Hindi, and Urdu. GitHub

The accurate description is:

SingChat MCP's public manifest, tool definitions, integration examples, and technical documentation have been released under the MIT License.

The public repository also explains that it stores the public manifest, examples, and documentation for the hosted SingChat MCP server. GitHub

It should therefore not currently be expanded into the claim:

All source code for every internal implementation of the hosted SingChat MCP server has been published.

Layer two: SingChat Community Edition

Official SingChat documentation describes Community Edition as:

  • licensed under AGPL-3.0;

  • free;

  • self-hostable;

  • runnable with Docker;

  • intended for users who want to control their own infrastructure and data. SingChat

Community Edition means an organization may choose to:

  • operate its own servers;

  • operate its own database;

  • manage its own attachment storage;

  • use its own AI model or API Key;

  • modify and deploy according to the license;

  • avoid being limited to the official SingChat SaaS.

The exact download, source-release scope, and version content of Community Edition should follow SingChat's official self-hosting documentation and release pages.

Layer three: SingChat Enterprise Edition

Official SingChat documentation positions Enterprise Edition as a commercially licensed edition.

Community Edition is made available for free under AGPL-3.0; Enterprise Edition enables the corresponding enterprise capabilities through a valid enterprise license. SingChat

This model allows SingChat to provide, at the same time:

  • public MCP integration material;

  • a free, self-hosted Community Edition;

  • the official hosted SaaS;

  • commercially licensed enterprise features and support.

Different components use different licenses. SingChat as a whole must not be described as being licensed uniformly under the MIT License.

7. What is SingChat MCP?

MCP stands for Model Context Protocol.

It lets AI assistants do more than generate text: within an authorized scope, they can use tools supplied by an external service.

Without MCP, a user normally has to:

  1. open the SingChat administration panel;

  2. create a workspace;

  3. configure a website;

  4. copy the embed code;

  5. add knowledge;

  6. configure the AI;

  7. review conversations;

  8. reply to visitors manually.

With SingChat MCP, these operations can be completed through an AI assistant.

For example, a user can tell Claude Code or Cursor:

Set up SingChat customer support for my website.

The AI assistant can run singchat_setup, create a free workspace, and return initial administration sign-in information, website embed code, and an API Key. GitHub

SingChat MCP turns the traditional process of operating a support administration panel page by page into a process in which a user asks AI, in natural language, to complete support setup.

8. Which tools does SingChat MCP currently provide?

SingChat MCP currently publishes 17 tools. GitHub

Workspace creation and claiming

singchat_setup

Creates a free workspace and returns initial sign-in information, website embed code, and an API Key.

claim_account

Returns the claim process that formally binds an AI-created workspace to the user's account.

get_embed_snippet

Returns website embed code for HTML, React, or Next.js.

Conversation management

list_conversations

Lists visitor conversations in the workspace.

get_conversation

Reads messages and status for a specified conversation.

reply_to_conversation

Replies directly to a visitor.

resolve_conversation

Marks a completed conversation as resolved.

AI knowledge base

add_knowledge

Adds knowledge by crawling a website or importing question-and-answer material.

list_knowledge

Shows imported knowledge and processing status.

search_knowledge

Searches the workspace knowledge base.

AI support configuration

configure_agent

Adjusts the AI model, prompt, and customer-support behavior.

set_keywords

Configures keywords and automatic replies.

manage_faq

Adds, updates, or deletes FAQs.

Analytics

get_analytics

Shows conversation volume, message volume, resolution rate, and other analytics.

AI knowledge connector

search

Searches workspace knowledge in a compatible format.

fetch

Retrieves specified knowledge content.

ping

Checks MCP service health.

These tools let developers complete part of the support configuration and management workflow inside an editor or AI chat interface rather than switching to the SingChat administration panel every time. GitHub

9. How do MCP tools restrict workspace permissions?

A customer-support system contains organizational knowledge, visitor conversations, and workspace settings. AI tools therefore cannot be given unrestricted cross-tenant access.

The SingChat MCP public documentation states:

Each tool acts only on the workspace bound to the API Key and does not accept an arbitrary tenant parameter. GitHub

The public design can be understood as:

User authorization or API Key
        ↓
Bound to a specified SingChat workspace
        ↓
MCP tools operate only in that workspace
        ↓
They cannot arbitrarily specify another tenant

Publishing this tool boundary is an important reason SingChat forms part of the SingLinkVPN transparency programme.

From the public documentation, external developers can understand:

  • tool names;

  • operations the tools can perform;

  • whether a tool touches conversations or knowledge;

  • how a workspace is bound to authorization credentials;

  • what AI can do;

  • what AI cannot arbitrarily specify.

Public tool restrictions do not, however, mean that users can ignore credential security.

API Keys, OAuth authorization, and MCP clients still require proper protection. They must not be included in public code, screenshots, or conversation content.

10. Which AI tools does SingChat MCP support?

The public documentation currently provides or explains integration methods for:

  • Claude Code;

  • Claude Desktop;

  • Claude on the web;

  • Cursor;

  • ChatGPT desktop;

  • ChatGPT on the web;

  • Grok;

  • Gemini CLI;

  • other tools supporting remote MCP;

  • clients supporting only stdio through a bridge. GitHub

All supported tools use the same SingChat MCP endpoint.

The public documentation also notes that the consumer Gemini App currently has no direct button for adding MCP, so Gemini CLI is the primary recommended integration. GitHub

MCP capabilities, account permissions, and configuration entry points in different AI products can change with product releases. Actual integration methods should follow the latest documentation for both the relevant AI product and SingChat.

11. How does SingChat AI customer support work?

SingChat's AI support answers questions from an organization's knowledge base.

The basic process is:

A visitor asks a question
        ↓
Search the organization's knowledge base
        ↓
Retrieve relevant documents or FAQs
        ↓
AI generates an answer from the knowledge
        ↓
Transfer to a human when AI cannot handle the case

The official SingChat website describes AI capabilities including:

  • streaming answers generated from the knowledge base;

  • support for an organization's own OpenAI or Anthropic model key;

  • keyword-based automatic replies;

  • one-click transfer to a human support agent. SingChat

For SingLinkVPN, the knowledge base can organize:

  • download and installation instructions;

  • plan and data rules;

  • client-update announcements;

  • node instructions;

  • common connection problems;

  • system-compatibility information;

  • security and privacy information;

  • after-sales handling procedures.

AI is suitable for first handling standardized, highly repetitive questions.

Complex accounts, payment disputes, unusual technical failures, and situations requiring human judgement should be handed to a human.

12. AI customer support does not mean removing human support

SingChat is not intended to replace human customer support entirely with AI.

AI is better suited to:

  • frequently asked questions;

  • product-documentation enquiries;

  • plan and policy questions;

  • basic usage instructions;

  • known issues and handling steps;

  • initial replies outside business hours.

Human agents are better suited to:

  • individual account and order cases;

  • unusual payment problems;

  • technical failures that cannot be reproduced;

  • complaints and disputes;

  • situations requiring judgement and coordination;

  • questions for which AI lacks sufficient information.

SingChat's official capabilities include both AI answers and human takeover. A human agent can review the conversation context in the shared inbox and continue handling it. SingChat

SingChat therefore follows this model:

AI first handles questions that can be standardized; humans take responsibility for cases requiring judgement, accountability, and further follow-up.

13. How do websites and apps integrate SingChat?

Website integration

A website can add the SingChat chat widget with one script.

The SingChat MCP get_embed_snippet tool returns embed examples for HTML, React, and Next.js. GitHub

App integration

The official SingChat website positions the product for embedding in:

  • Web;

  • iOS;

  • Android;

  • Flutter;

and other platforms. SingChat

For SingLinkVPN, this means that support entry points can progressively be integrated into:

  • the official website;

  • the Mac client;

  • the Windows client;

  • mobile apps;

  • the user dashboard;

  • other product interfaces.

Different platforms can use the website widget, WebView, APIs, or another corresponding integration. Actual capabilities should follow the latest SingChat developer documentation.

14. SingChat's multi-tenant and security design

SingChat must serve different organizations and workspaces at the same time. Tenant isolation is therefore a fundamental requirement.

Security and permission capabilities listed by the official website include:

  • PostgreSQL Row-Level Security;

  • team roles and fine-grained permissions;

  • scoped API Keys;

  • separate branding and domains for each workspace;

  • webhooks;

  • SSO capabilities in Enterprise Edition. SingChat

Multi-tenant isolation cannot depend only on hiding buttons in the front end.

A support system must restrict permissions at multiple levels:

  • accounts and workspaces;

  • database queries;

  • API Keys;

  • MCP tools;

  • webhooks;

  • attachments;

  • knowledge bases;

  • conversations;

  • agent roles.

Publishing MCP tool definitions and the method used to bind a workspace does not mean that SingChat has been proven free of every vulnerability.

It means external developers can begin to understand and inspect the public integration layer. Security testing, vulnerability disclosure, permission review, and version updates remain necessary.

15. How does SingChat handle SingLinkVPN support data?

When SingChat acts as SingLinkVPN's support system, it processes support and after-sales data voluntarily submitted by users, not browsing-activity logs automatically created in the background by a VPN.

Material voluntarily submitted by users may include:

  • a problem description;

  • client version;

  • device and operating-system version;

  • error information;

  • screenshots;

  • order-enquiry information;

  • diagnostic information supplied by the user;

  • customer-support conversation content.

This material is different from VPN activity data such as:

  • websites visited by the user;

  • destination domain names;

  • complete URLs;

  • DNS queries;

  • source-IP history;

  • browsing content;

  • network-communication content.

As SingChat becomes part of the SingLinkVPN transparency programme, future disclosure and improvement should continue to cover:

  • purposes for which support data is used;

  • how conversations and tickets are deleted;

  • whether remote diagnostics are submitted voluntarily by users;

  • permissions for different support roles;

  • the authorization scope of APIs and MCP;

  • handling of attachments and knowledge bases;

  • how security issues can be reported.

Open-source integration documentation cannot replace a complete privacy policy, but it can make the system's tools and permission design easier for outsiders to understand.

16. SingChat Community Edition and the official SaaS

SingChat offers different usage models.

Official SingChat SaaS

SingChat provides and maintains:

  • servers;

  • databases;

  • real-time communication;

  • AI services;

  • storage;

  • updates;

  • fundamental operations and maintenance.

The official website offers a free plan and lists plan content including one support agent, 100 AI answers per month, five knowledge documents, and the website widget. SingChat

Community Edition

Community Edition is deployed by the user.

Official documentation identifies Community Edition as AGPL-3.0 and says that it can run on the user's own server. SingChat

It is suited to organizations requiring:

  • their own database;

  • deployment in a designated region;

  • their own attachment storage;

  • their own AI model;

  • internal-network integration;

  • more complete control over data.

Enterprise Edition

Enterprise Edition uses a commercial license and is intended for organizations that need enterprise capabilities, licensing, and commercial support. SingChat

17. What is the difference between MIT and AGPL?

Different SingChat components use different licenses.

Public SingChat MCP repository: MIT License

MIT is a permissive open-source license. It generally permits use, copying, modification, publication, distribution, and commercial use, provided that the corresponding copyright and license notices are retained.

At present, MIT applies to the material actually published in the singchat-mcp repository. GitHub

Community Edition: AGPL-3.0

AGPL is an open-source license that includes source-sharing requirements for network services.

When a user modifies AGPL software and provides a service to others over a network, the corresponding source-offer obligations must be handled according to the AGPL terms.

Enterprise Edition: commercial license

Enterprise Edition is governed by a commercial license for enterprise capabilities, commercial deployment, and corresponding support.

The SingChat licensing structure should therefore be described accurately as:

Public SingChat MCP repository → MIT
Community Edition → AGPL-3.0
Enterprise Edition → Commercial license

It must not be described as:

All SingChat products and all server-side source code use the MIT License.

18. Open source does not mean free hosting

Open-source software and an officially hosted service are two different things.

Open source means users may, according to the applicable license:

  • inspect the published material;

  • use it;

  • modify it;

  • self-host it;

  • submit improvements;

  • build integrations.

Official hosting means SingChat must provide:

  • servers;

  • databases;

  • AI compute;

  • real-time communication;

  • attachment storage;

  • backups;

  • updates;

  • operations and maintenance.

Therefore, even if Community Edition can be self-hosted without a license fee, the official SaaS can still offer different plans based on the number of agents, AI usage, knowledge-base capacity, enterprise features, and service levels. The official website offers both free and paid hosted plans. SingChat

19. Open source does not mean unrestricted use of the SingChat brand

Software licensing and trademark rights are not the same.

MIT and AGPL govern rights to use the corresponding public code and documents. They do not automatically grant anyone:

  • the SingChat name;

  • the SingChat Logo;

  • official website domains;

  • official verified status;

  • an official commercial relationship.

Developers may use and modify public material according to the relevant license, but when launching a modified version as an independent service they must still comply with applicable branding and trademark rules.

“May use open-source code” must therefore not be interpreted as:

May operate another identical service under the official SingChat brand.

20. Why is SingLinkVPN publishing SingChat?

Publishing SingChat is part of the extension of the overall SingLinkVPN open-source and transparency programme into customer-support systems.

There are five principal reasons.

1. Make AI customer-support capabilities understandable

Instead of making a vague claim that “AI can manage customer support,” SingChat publishes specific tools, operations, and workspace restrictions.

2. Make third-party integration methods inspectable

Developers can review how Claude, Cursor, ChatGPT, Grok, and Gemini CLI integrate with SingChat. GitHub

3. Make licensing boundaries clearer

The programme explicitly distinguishes:

  • the MIT-licensed public MCP repository;

  • AGPL Community Edition;

  • commercially licensed Enterprise Edition.

4. Lower integration costs

Developers can add customer support directly through MCP, APIs, or website embedding instead of building an entire chat, knowledge-base, and human-takeover system from scratch.

5. Accept community review and improvement

The community can:

  • review public documentation;

  • test MCP integrations;

  • submit issues;

  • correct multilingual material;

  • add integration examples;

  • report security problems;

  • suggest new tools.

Transparency does not mean publishing all sensitive production configuration. It means disclosing enough information for outsiders to understand, inspect, and verify the system without harming user privacy or system security.

21. How should SingChat's current public scope be described?

For formal public communication, the recommended description is:

SingChat is the customer-support and AI assistance system currently used by SingLinkVPN and is part of the SingLinkVPN open-source and transparency programme. SingChat MCP's public manifest, tool descriptions, integration examples, and multilingual documentation have been published under the MIT License.

Another accurate description is:

SingLinkVPN is extending open source and transparency from VPN protocols, clients, and security systems to user-support systems. As the first phase, the SingChat MCP integration layer and public technical documentation are now open.

The following descriptions are not recommended:

  • all SingChat source code has been fully open sourced;

  • every SingChat component uses the MIT License;

  • all code for the hosted SingChat server is public;

  • after the open-source release, anyone may directly use the official SingChat brand;

  • using SingChat means that no customer-support data is processed;

  • publishing MCP documentation is equivalent to completing a comprehensive security audit.

22. How can the community participate?

The community can currently participate through the public SingChat MCP repository by:

  • submitting issues;

  • testing different AI clients;

  • improving integration documentation;

  • correcting multilingual translations;

  • adding website-embedding examples;

  • proposing new MCP tools;

  • reporting permission or security issues;

  • improving self-hosting documentation;

  • submitting product and compatibility feedback.

Future public engineering work should progressively improve:

  • CONTRIBUTING.md;

  • SECURITY.md;

  • CODE_OF_CONDUCT.md;

  • CHANGELOG.md;

  • the version-support policy;

  • the vulnerability-disclosure process;

  • release checksums;

  • Community Edition release records;

  • reproducible deployment tests;

  • permission and data-processing documentation.

23. SingChat's future disclosure plan

The SingChat open-source and transparency programme can be divided into three phases.

Phase one: MCP integration layer

The programme has begun publishing:

  • the public MCP manifest;

  • tool descriptions;

  • multi-platform integration examples;

  • multilingual README files;

  • the MIT License;

  • workspace-creation methods;

  • conversation and knowledge-management tools;

  • workspace permission boundaries.

Phase two: Community Edition and deployment information

Future improvements include:

  • the official Community Edition release entry point;

  • deployment documentation;

  • Docker configuration;

  • environment-variable examples;

  • version tags;

  • update and upgrade instructions;

  • backup and recovery guides;

  • a security policy;

  • data-processing and retention information.

Phase three: community verification and ecosystem

Future development may include:

  • additional MCP tools;

  • additional AI models;

  • additional messaging channels;

  • additional app-integration components;

  • community plug-ins;

  • security testing;

  • third-party evaluation;

  • a public Roadmap;

  • reproducible deployment and performance tests;

  • multilingual community contributions.

The SingChat open-source programme starts with MCP, but its public scope should not remain limited to MCP forever.

24. Frequently asked questions

What is SingChat?

SingChat is a customer-support platform combining AI support, human agents, a shared inbox, a RAG knowledge base, a website widget, APIs, webhooks, MCP, and self-hosting. SingChat

SingChat is the live support and AI assistance system currently used by SingLinkVPN and is also part of the SingLinkVPN open-source and transparency programme.

Why should a VPN disclose its support system?

Because data interaction between users and a VPN service does not occur only inside the VPN tunnel. Support conversations, account enquiries, device information, and diagnostics voluntarily submitted by users also require clear purpose, permission, and technical boundaries.

Which parts of SingChat are currently public?

The singchat-mcp repository currently publishes the MCP manifest, integration examples, multilingual documentation, tool descriptions, and the MIT License. GitHub

Is the complete SingChat MCP server open source?

The public repository says that it contains the public manifest, examples, and documentation for the hosted SingChat MCP server. It therefore must not be described as publishing the source code for every hosted server implementation. GitHub

Which license does SingChat MCP use?

The public singchat-mcp repository uses the MIT License. GitHub

Which license does SingChat Community Edition use?

Official SingChat documentation identifies Community Edition as AGPL-3.0. SingChat

Can SingChat be self-hosted?

The official website and documentation position Community Edition as a free, self-hostable edition. SingChat

How many tools does SingChat MCP provide?

The public documentation currently lists 17 tools covering workspace creation, website embedding, conversation management, knowledge, AI configuration, FAQs, and analytics. GitHub

Does SingChat support ChatGPT?

The public documentation provides instructions for ChatGPT remote MCP and the search and fetch knowledge connector. Actual availability depends on the user's ChatGPT account and current product version. GitHub

Does SingChat support Claude and Cursor?

Yes. The public documentation provides integration methods for Claude Code and Cursor. GitHub

Does SingChat support Gemini?

The public documentation currently primarily recommends Gemini CLI. The consumer Gemini App does not presently provide a direct button for adding MCP. GitHub

Does SingChat process no customer-support data at all?

No. A customer-support system must process conversations, problem descriptions, and after-sales information voluntarily submitted by users. This is not the same category as a VPN automatically logging websites, DNS, or network activity.

Does open source mean that the official SingChat SaaS is free?

No. Open-source licensing and an officially hosted service are different models. Community Edition can be self-hosted under its license, while the official SaaS supplies servers, updates, AI resources, and operations. SingChat

Can the SingChat brand be used directly after an open-source release?

An open-source code license does not automatically grant trademark rights to the SingChat name and Logo.

Conclusion

Opening SingChat is more than a technical release for an independent AI customer-support product.

It is also part of the SingLinkVPN open-source and transparency programme.

SingLinkVPN transparency should not be limited to VPN protocols, clients, security models, the no-logs policy, and performance data. Customer-support entry points, AI tools, knowledge bases, workspace permissions, and handling of after-sales data between users and the product must also be progressively disclosed and subjected to review.

SingChat is the customer-support and AI assistance system currently used by SingLinkVPN.

It handles product, account, node, connection, compatibility, and technical-support questions submitted by users and assists with AI answers, knowledge-base retrieval, and human takeover.

At present, SingChat MCP's public manifest, tool descriptions, integration examples, and multilingual documentation have been published under the MIT License. The public documentation lists 17 MCP tools that let MCP-capable AI assistants create a workspace, add website support, manage conversations, train AI, and review analytics. GitHub

The official SingChat website also provides cross-platform AI support, human takeover, a shared inbox, multi-tenant isolation, APIs, webhooks, and a choice between the official SaaS and self-hosted AGPL Community Edition. SingChat

Open-source claims must, however, remain accurate.

The public singchat-mcp repository currently contains primarily the public manifest, examples, and documentation for the hosted MCP service. It must not be described as publishing all hosted server source code. Additional support-system, deployment, security, and Community Edition material will be improved in phases according to security, privacy, licensing, and product stability.

The transparency SingLinkVPN aims to establish is not merely promotional visibility. It is intended to let the community progressively understand:

  • how the VPN transports data;

  • which limited data the system retains;

  • how the no-logs policy is verified;

  • how performance conclusions are tested;

  • which operations AI customer support can perform;

  • how support-workspace permissions are restricted;

  • how after-sales data voluntarily submitted by users is handled;

  • which content is already public;

  • which material is still under review and preparation.

Opening SingChat shows that the SingLinkVPN open-source and transparency programme is extending from underlying VPN technology into the customer-support and assistance systems that users actually interact with.

It starts with MCP, but it will not stop at MCP.

Sources