slideless-mcp
MCP server that wraps the Slideless HTTP API as tools for listing, sharing, uploading, and managing HTML presentations from any MCP host without installing the CLI.
README
slideless-mcp
MCP server for Slideless. Wraps the Slideless HTTP API as Model Context Protocol tools so users can list, share, upload, and manage HTML presentations from any MCP host (Claude Desktop, claude.ai, ChatGPT desktop, Cursor, etc.) without installing the slideless CLI.
Architecture
Stateless Cloudflare Worker (Durable Object per session for MCP state). The Worker forwards the user's Authorization: Bearer cko_… header to the Slideless Cloud Functions in europe-west1. No database, no secrets to rotate — the user's API key never leaves the connector header.
Claude / ChatGPT → mcp.slideless.ai/mcp (Cloudflare Worker)
│
▼
europe-west1-slideless-ai.cloudfunctions.net
Use as a connector
- Sign in at https://app.slideless.ai → Settings → API Keys → Create
- Copy the
cko_…key (shown once) - In Claude Desktop: Settings → Connectors → Add custom connector
- URL:
https://mcp.slideless.ai/mcp - Header:
Authorization: Bearer cko_…
- URL:
- Connect. The server validates the key on connect; bad keys fail fast.
Tools
| Tool | What it does |
|---|---|
slideless_whoami |
Identity check — returns org, key name, scopes |
slideless_list_presentations |
List owned + invited presentations |
slideless_get_presentation |
Full info for one presentation (tokens, collaborators) |
slideless_list_versions |
Version history of a presentation |
slideless_get_version |
Manifest of a specific version |
slideless_download_version |
Manifest + inline text-file contents (HTML/CSS/JS up to 256 KB each) |
slideless_upload_html_presentation |
Upload a single-file HTML deck |
slideless_upload_presentation_files |
Upload a multi-file deck (base64 array) |
slideless_add_share_token |
Mint a public viewer URL |
slideless_set_token_version_mode |
Pin or unpin a token to a version |
slideless_unshare_presentation |
Revoke one or all share tokens |
slideless_share_via_email |
Send the share URL by email |
slideless_invite_collaborator |
Grant another user dev access (sends email) |
slideless_uninvite_collaborator |
Revoke a collaborator |
slideless_list_collaborators |
List collaborators on a presentation |
slideless_delete_presentation |
Permanently delete a presentation |
Local development
npm install
npm run dev # wrangler dev → http://localhost:8787
Test with the MCP Inspector:
npx @modelcontextprotocol/inspector
# Set transport: HTTP (streamable)
# URL: http://localhost:8787/mcp
# Header: Authorization: Bearer cko_<your-key>
Deploy
npm run deploy
The first deploy creates the Worker at https://slideless-mcp.<account>.workers.dev/mcp. Bind the custom domain mcp.slideless.ai via Cloudflare dashboard or wrangler deploy --routes once DNS is in place.
Type checking:
npm run typecheck
Source layout
src/
├── index.ts # Worker entry: routing, rate limiting, McpAgent
├── server.ts # Tool registration entry point
├── slidelessClient.ts # Typed fetch wrapper around Cloud Functions
├── types.ts # Wire shapes (mirrors slideless-app types/)
├── errors.ts # SlidelessApiError + wrapToolErrors helper
└── tools/
├── identity.ts # slideless_whoami
├── presentations.ts # list / get / versions / download / delete
├── upload.ts # upload_html / upload_files (3-step orchestration)
├── sharing.ts # tokens, version mode, unshare, email
└── collaborators.ts # invite / uninvite / list
Related repos
slideless-app— backend Cloud Functions this server proxiesslideless-cli— npm CLI that uses the same APIslideless-plugin— Claude Code plugin (companion authoring + upload skills)
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。