Loading...
A client-side web GUI to manage and configure Model Context Protocol (MCP) servers for the Claude Desktop application, simplifying setup and access to various services.
Boost this tool
Subscribe to listing upgrades or segmented pushes.
A client-side web GUI to manage and configure Model Context Protocol (MCP) servers for the Claude Desktop application, simplifying setup and access to various services.
The MCP Manager itself is relatively safe due to its client-side nature. However, the overall security depends heavily on the configuration and security of the managed MCP servers. It is safe to use with read-only servers or well-sandboxed environments. It becomes risky when managing servers with write access to sensitive data or systems without proper access controls.
Performance depends on the underlying MCP servers and the network connection. Some servers may have rate limits or other performance constraints.
Cost depends on the APIs and services used by the MCP servers. Consider API usage costs, token consumption, and resource utilization.
Apple NotesAccesses and searches your Apple Notes.
Primarily read-only access, limited write capabilities.
AWS Knowledge BaseAccesses and queries AWS Knowledge Base for information retrieval.
Read-only access to knowledge base content.
Brave SearchSearches the web using the Brave Search API.
Read-only access to search results.
BrowserbaseEnables Claude to explore the web with Browserbase.
Allows web browsing, which could lead to unintended actions or information exposure.
CloudflareManages Cloudflare workers and account resources.
Allows modification of Cloudflare configurations, potentially impacting website availability and security.
EverartInterfaces with Everart API for digital art and design tools.
Depends on the specific Everart API capabilities; write operations could modify art assets.
ExaSearches the web with Exa.
Read-only access to search results.
FilesystemAccesses and manages the local filesystem.
Allows arbitrary file read/write/delete operations.
GitHubAccesses your GitHub repositories.
Allows code reading and potentially writing, depending on permissions.
GitLabManages GitLab repositories and resources.
Allows modification of GitLab configurations and code, potentially impacting project integrity.
Google DriveAccesses and searches files in your Google Drive.
Allows read and potentially write access to files in Google Drive.
Google MapsAccesses Google Maps API for location services.
Primarily read-only access to location data.
MemoryGives Claude memory of previous conversations.
Stores conversation history, posing minimal risk.
ObsidianReads and searches files in your Obsidian vault.
Primarily read-only access to Obsidian notes.
PerplexitySearches the web with Perplexity API.
Read-only access to search results.
PostgreSQLConnects and interacts with PostgreSQL databases.
Allows arbitrary SQL queries, potentially modifying or deleting data.
PuppeteerAutomates browser interactions.
Allows arbitrary browser actions, potentially leading to unintended consequences.
Sequential ThinkingEnables step-by-step reasoning.
Enhances reasoning capabilities without direct access to external resources.
SlackAccesses your Slack workspace.
Allows reading and potentially sending messages in Slack.
SQLiteManages SQLite databases.
Allows arbitrary SQL queries, potentially modifying or deleting data.
TodoistAccesses and searches your Todoist tasks.
Allows read and potentially write access to Todoist tasks.
YouTube TranscriptAccesses and searches YouTube transcripts.
Read-only access to YouTube transcript data.
Environment Variable
cloud
The MCP Manager itself is relatively safe due to its client-side nature. However, the overall security depends heavily on the configuration and security of the managed MCP servers. It is safe to use with read-only servers or well-sandboxed environments. It becomes risky when managing servers with write access to sensitive data or systems without proper access controls.
Autonomy depends entirely on the configuration of the individual MCP servers. Some servers may offer dry-run modes or rollback support, but this is not guaranteed.
Production Tip
Carefully configure and monitor MCP servers to ensure they have appropriate permissions and are not exposing sensitive data.
No, this project is not affiliated with Anthropic.
The MCP Manager itself runs client-side, but the security depends on the configuration of the managed MCP servers. Ensure servers have appropriate permissions and are not exposing sensitive data.
The MCP servers typically use environment variables for authentication, such as API keys or tokens.
Yes, contributions are welcome! You can open a PR with new MCP servers or any other improvements to the codebase.
No, this tool runs entirely client-side, so your data never leaves your computer.
The security of the MCP server is independent of this manager. Ensure the MCP server itself is secure and up-to-date.
Misconfiguring an MCP server can lead to unintended data exposure or unauthorized actions. Carefully review the configuration before applying it.