SumoSign MCP Server
Exposes the SumoSign signing API to AI agents, allowing document upload, envelope creation, and tracking, with signing always done by human recipients.
README
@sumosign/mcp
MCP server exposing the SumoSign signing API to AI agents (Claude, ChatGPT, Copilot, and any MCP-compatible client) over stdio.
The API key authenticates the agent, not a signer. It can never complete a signature — signing is always done by a human recipient through their emailed, one-time link, with consent capture and an append-only audit trail.
Install & run
No install step is required; run it directly with npx:
SUMOSIGN_API_KEY=ss_live_... npx @sumosign/mcp
Configuration
| Env var | Required | Default | Description |
|---|---|---|---|
SUMOSIGN_API_KEY |
yes | — | A scoped ss_live_... key from the SumoSign portal. |
SUMOSIGN_API_URL |
no | https://api.sumosign.app |
Override for self-hosted or staging APIs. |
Client config example
{
"mcpServers": {
"sumosign": {
"command": "npx",
"args": ["-y", "@sumosign/mcp"],
"env": { "SUMOSIGN_API_KEY": "ss_live_..." }
}
}
}
Tools
The server registers 15 tools:
Documents
upload_document— upload a local PDF; returns the document id used to create envelopes.analyze_document— analyze an uploaded PDF for field-placement suggestions (text extraction, no OCR).
Templates
create_template— create a reusable template from a PDF with text anchors like{{signature_1}}.list_templates— list the organization's reusable templates.seed_starter_templates— seed sample templates (test envelope, simple NDA, one-page agreement).
Envelopes — create
create_envelope— create a draft envelope from an uploaded document (inline fields) or a template.create_envelope_from_template— create a draft envelope from a template; fields are inherited.
Envelopes — verify & send
preview_envelope— download a placement preview (every field drawn as a dashed, labeled box) to a local path. No email is sent and no envelope is consumed — use it to confirm placement with a human before sending.send_envelope— send a draft envelope. Only humans can sign; this tool cannot complete a signature.
Envelopes — track
get_envelope— get envelope status including per-recipient progress.list_envelopes— list the most recent envelopes for the organization.void_envelope— void an envelope before completion; revokes outstanding signing links (audit-logged).get_audit_trail— get the append-only audit trail with actor attribution.
Downloads
download_signed_pdf— download the flattened, signed PDF of a completed envelope to a local path.download_certificate— download the Certificate of Completion to a local path.
Security
- Local filesystem access.
upload_documentreads a local PDF path you provide, andpreview_envelope,download_signed_pdf, anddownload_certificatewrite files to local paths you provide. The server can therefore read and write any file the host process can reach. Run it with least privilege and only point it at paths you intend to share. - Untrusted document content / prompt injection.
analyze_documentextracts text from PDFs and returns it to the agent. Document contents are untrusted input and may contain prompt-injection attempts. Treat extracted text as data, not instructions, and keep a human in the loop beforesend_envelope. - Least-privilege API keys. Issue a scoped key per agent from the SumoSign portal and grant
only the scopes that agent needs. Available scopes:
documents:create,documents:read,documents:send,documents:download,templates:read,templates:write,signers:manage,webhooks:write,audit_logs:read,branding:write,voice:trigger. Keys are hashed at rest and revocable, and can never complete a signature on a human's behalf.
About
SumoSign is developed by Cipher Projects.
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。