Synerixis Ops MCP
Synerixis Ops MCP exposes merchant inbox operations for Shopify/Amazon sellers, with read tools plus approval-gated write tools.
List connected stores and conversations, including filters by store, status, limit, and offset.
Search conversations by order number or buyer email, and inspect conversation details, messages, and linked order/logistics context.
View pending AI draft replies, AI resolve-point quota, and available agents/seats.
Save or update draft replies without sending.
Send replies only when
confirm=trueand the host approves the destructive call; retries can use an idempotency key.Approve and send one pending AI draft, or reject/discard one AI draft with a stored reason.
Assign conversations to an agent after listing agents, and update status to
resolved,reopen, orclosed.Write operations are marked destructive and require host approval; the Synerixis API audits attempts/results and rejects writes without confirmation.
Does not provide autosend, refunds, logistics writes, or secrets; those routes return 403.
Provides tools for Amazon sellers' customer-support inbox operations via the Synerixis Ops API: list and search conversations, read messages and order context, save drafts, and send replies (with host approval), plus assign conversations, update status, and manage pending AI replies.
Provides tools for Shopify sellers' customer-support inbox operations via the Synerixis Ops API: list and search conversations, read messages and order context, save drafts, and send replies (with host approval), plus assign conversations, update status, and manage pending AI replies.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Synerixis Ops MCP Search my inbox for order #1042 and show the conversation"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Synerixis Ops MCP
跨境电商 · 智能客服工作台(默认远程 HTTPS MCP)
Lightweight MCP shell for Synerixis merchant inbox ops.
极简 MCP 壳:不挂本地数据库,不包含 Synerixis 业务源码;所有数据经正式 API。
Default install = remote HTTPS → https://api.synerixis.com/api/ops-mcp/v1/mcp(Shopify 同款绿点 HTTP)。
本仓仍提供可选 stdio(.mcp.stdio.json / npx)给本机 Claude Desktop。
Agent / product docs: PRODUCT.md · USAGE.md · skills under skills/ (synerixis-ops-inbox, synerixis-ops-connect, synerixis-ops-pending-ai).
English
What this is
A thin Model Context Protocol surface that lets an external assistant list conversations, read order context, save drafts, and send replies after the host approves the tool call.
Heavy lifting stays on the Synerixis API:
https://api.synerixis.com/api/ops-mcp/v1
Write tools are marked destructiveHint. The host must approve destructive calls. The Synerixis API still rejects a write unless confirm is true, and it writes an audit row for the attempt and the result.
Docs for Grok / Cursor skills: PRODUCT.md, USAGE.md, skills/synerixis-ops-inbox.
What this is not
Not the Synerixis main product repo
No local DB / Redis / queue
No marketplace plugin install (Directory listing is separate)
No autosend, refunds, logistics writes, or secrets. The API returns 403 for those routes.
Who this is for (beta)
Independent Shopify sellers and Amazon sellers running customer support through Synerixis. Public beta — still testing. Expect rough edges; feedback: support@synerixis.com.
Install (recommended) — remote HTTPS
Root mcp.json / .mcp.json already default to remote. After install, fill the sxk_ key (plugin variables card, Grok secure card, or headers).
{
"mcpServers": {
"synerixis-ops": {
"url": "https://api.synerixis.com/api/ops-mcp/v1/mcp",
"headers": {
"Authorization": "Bearer ${SYNERIXIS_OPS_MCP_API_KEY}"
}
}
}
}Generate a key in Synerixis merchant console → Connections → enable external assistant → Generate (
sxk_…).Install from cursor.directory, this GitHub repo, or paste the URL into Grok Bot / Cursor MCP settings.
Set
SYNERIXIS_OPS_MCP_API_KEY(or pasteBearer sxk_…into headers). Never commit a real key.Connected HTTP transport shows a green status light (same class as Shopify’s remote MCP).
Auth: Authorization: Bearer sxk_… (or X-Api-Key). v1 is stateless JSON (no SSE session).
Do not put the key in the URL or in this repo.
Manual Cursor user config can use ${env:SYNERIXIS_OPS_MCP_API_KEY} instead of the plugin ${SYNERIXIS_OPS_MCP_API_KEY} placeholder. See REMOTE.md and AUTH.md.
Optional: stdio (local Claude Desktop / Cursor)
Only if you need a local Node process. Copy .mcp.stdio.json and replace sxk_xxxxxxxx with the real key. ${SYNERIXIS_OPS_MCP_API_KEY} is only expanded in the plugin mcp.json, not in this stdio file. Cursor manual config can use ${env:SYNERIXIS_OPS_MCP_API_KEY}.
npx -y github:sdrom2008/synerixis-ops-mcp{
"mcpServers": {
"synerixis-ops": {
"command": "npx",
"args": ["-y", "github:sdrom2008/synerixis-ops-mcp"],
"env": {
"SYNERIXIS_OPS_MCP_API_KEY": "sxk_xxxxxxxx"
}
}
}
}Or clone + npm install + npm run build and point command at dist/index.js.
The process reads only SYNERIXIS_OPS_MCP_API_KEY. It does not fall back to SYNERIXIS_API_KEY.
Optional SYNERIXIS_OPS_MCP_BASE_URL must be https on api.synerixis.com. A local API needs SYNERIXIS_OPS_MCP_ALLOW_INSECURE_LOCAL=1 and http://localhost or http://127.0.0.1.
Optional SYNERIXIS_OPS_MCP_ALLOWED_HOST_SUFFIX defaults to api.synerixis.com and must itself be synerixis.com or a subdomain of it. It cannot point the API key at another domain.
Tools
Tool | Purpose |
| Connected stores |
| Inbox list ( |
| Search by order number or buyer email ( |
| Conversation detail |
| Messages ( |
| Order context for this conversation's platform and store |
| AI drafts awaiting confirm |
| AI resolve points |
| Tier / expiry / renew portal (read-only; no checkout) |
| Agents/seats (id, display name, available) for assign |
| Save draft (no send) |
| Send. Host approval + |
| Send one AI draft ( |
| Discard one AI draft and store |
| Assign. Body sends |
|
|
Suggested flow: search_conversations or list_conversations → list_messages / get_order_context → save_draft; call list_agents before assign_conversation.
Auth
See AUTH.md. Keys look like sxk_ plus hex. This repo never stores tenant secrets.
License
MIT
Related MCP server: Summer MCP
中文
这是什么
面向商家的轻量 MCP(默认远程 HTTPS,可选 stdio)。外部助手可以列会话、看订单、存草稿,并在宿主批准这次工具调用之后发送。
重活仍在正式 API:https://api.synerixis.com/api/ops-mcp/v1
本仓 无本地数据库;默认装法连 api.synerixis.com 的 Streamable HTTP,不跑本机 Node。
写工具带 destructiveHint。宿主须批准破坏性调用。API 仍拒绝 confirm 不是 true 的写操作,并记下审计。
助手自学文档:PRODUCT.md、USAGE.md、skills/。
给谁用(内测)
面向用 Synerixis 做客服的 Shopify / Amazon 独立卖家。目前测试中,功能还在打磨;问题请发 support@synerixis.com。
安装(推荐)— 远程 HTTPS
根目录 mcp.json / .mcp.json 已默认远程。装完后填 sxk_ Key(插件变量卡 / Grok 安全卡 / headers)。
商家台 连接 页打开「允许外部助手」并 生成 Key。
从 cursor.directory、本仓库或 Grok Bot / Cursor MCP 设置安装。
填入
SYNERIXIS_OPS_MCP_API_KEY(或Bearer sxk_…)。禁止把真实 Key 提交进 Git。HTTP 连上后状态灯为绿色(与 Shopify 远程 MCP 同类)。
鉴权:Authorization: Bearer sxk_…。v1 无状态 JSON(无 SSE Session)。
详见 REMOTE.md、AUTH.md。
可选:本地 stdio
见上方 English。复制 .mcp.stdio.json 后,把 sxk_xxxxxxxx 换成真实 Key。${SYNERIXIS_OPS_MCP_API_KEY} 只在插件 mcp.json 里展开。环境变量只用 SYNERIXIS_OPS_MCP_API_KEY(sxk_ 开头)。
硬禁
API 对这些路径返回 403:自动发送、退款、物流写入、读取密钥。
鉴权
见 AUTH.md。
开源协议
MIT
Available Tools
15 toolsapprove_ai_replyBDestructive
Approve one pending AI draft and send it. Pass draftId when more than one draft is waiting. Host must approve. confirm is not set by this client.
Inputs: confirm (required), conversationId (required), draftId, idempotencyKey.
assign_conversationADestructive
Assign a conversation to an agent. Call list_agents first for assigneeId. 先 list_agents 取坐席 id。Sends agentId (assigneeId is the same value). Host must approve.
Inputs: confirm (required), assigneeId (required), conversationId (required).
get_conversationCRead-only
Get one conversation detail by id.
Inputs: conversationId (required).
get_order_contextCRead-only
Get order / logistics context linked to a conversation (read-only).
Inputs: conversationId (required).
get_quotaARead-only
Get AI resolve-point balance for this merchant.
Inputs: no named parameters.
list_agentsARead-only
List this merchant's agents/seats (id, displayName, isActive/isAvailable) for assign_conversation. 列出本店坐席(id、显示名、是否可用),供分派会话选用。Does not expose secrets or emails.
Inputs: no named parameters.
list_conversationsBRead-only
List inbox conversations. Optional filters: storeId, status, limit, offset. storeId is a connection id or the store open id. offset is a non-negative page offset.
Inputs: limit, offset, status, storeId.
list_messagesCRead-only
List messages in a conversation. offset skips from the newest page.
Inputs: conversationId (required), limit, offset.
list_pending_ai_repliesCRead-only
List AI draft replies awaiting confirm before send.
Inputs: limit, offset, storeId.
list_storesARead-only
List connected stores for this merchant tenant.
Inputs: no named parameters.
reject_ai_replyBDestructive
Discard one pending AI draft. Does not discard other drafts. Host must approve. Reason is stored on the audit row.
Inputs: confirm (required), conversationId (required), reason, draftId.
save_draftAIdempotent
Save or update a draft reply on a conversation. Does not send.
Inputs: content (required), conversationId (required).
search_conversationsARead-only
Search this merchant's conversations by order number or buyer email (q). Optional storeId/limit/offset. 按订单号或买家邮箱搜本店会话。Same summary fields as list_conversations; no secrets.
Inputs: q (required), limit, offset, storeId.
send_messageADestructive
Send a reply. Destructive: the host must approve the call. confirm must already be true; this client never sets it. Audited on Synerixis. Same idempotencyKey retries the same send.
Inputs: confirm (required), content (required), conversationId (required), idempotencyKey.
update_conversation_statusADestructive
Set conversation status to resolved, reopen, or closed. Host must approve. open is not a write value.
Inputs: status (required), confirm (required), conversationId (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 15 tools
Most tools have clearly distinct purposes: conversation listing/reading, message sending, AI draft approval/rejection, assignment, and status updates are separable. The main overlap is list_conversations vs search_conversations, which both return similar summary fields and differ mainly by the optional q filter, so an agent could occasionally choose the wrong one.
All tool names use consistent snake_case with predictable verb_noun structure, such as list_conversations, get_conversation, save_draft, send_message, and approve_ai_reply. No mixed conventions or vague names are present.
Fifteen tools is well-scoped for an operations inbox server covering stores, conversations, messages, AI drafts, agents, quota, and order context. Each tool appears to earn its place without excessive surface area.
The surface covers the core lifecycle: list/search/read conversations, read messages, inspect order context, save/send replies, manage AI drafts, assign conversations, and update status. Minor gaps exist around broader agent/store management and possible read-state handling, but these are not fatal for the stated ops purpose.
Maintenance
Related MCP Connectors
Hosted support inbox over MCP: answer web chat, WhatsApp, email and SMS, manage knowledge.
Work your miniDesk support inbox from AI apps: read conversations, search docs, draft replies.
Read and answer support conversations, and run outbound email and chat sequences, from your AI.
WhatsApp Business inbox for AI agents: read chats, reply, send approved templates.
Related MCP Servers
AlicenseNot gradedqualityBmaintenanceEnables AI assistants to find and read customer conversations, look up profiles, send replies, manage workspace metadata, and handle daily support workflows. It also supports triage, drafting, escalation, and handover tasks through MCP tools, prompts, and resources.72 npm1MIT
Summer MCPofficial
AlicenseNot gradedqualityBmaintenanceEnables AI assistants to investigate stuck orders and billing issues, read customer conversations, connect telehealth support tools, review evidence, and prepare proposed actions for human approval through a hosted, OAuth-protected MCP endpoint.MIT- AlicenseAqualityBmaintenanceEnables AI assistants to read and manage support conversations, including attachments and local audio transcription, and to draft, approve, send, transfer, or conclude chats with human confirmation.371MIT
- AlicenseNot gradedqualityCmaintenanceEnables AI assistants to ask support agents questions, manage conversations and leads, and keep knowledge sources current through a hosted, OAuth-secured connection.MIT