ThinkWatch
by ThinkWatchProject·★ 868·Score 54
Enterprise AI bastion host with unified proxying, RBAC, audit logs, rate limiting, and cost tracking across multiple LLM providers and MCP servers.
Overview
ThinkWatch is an enterprise-grade security gateway for AI APIs and MCP (Model Context Protocol) access. It provides a unified control plane that authenticates, authorizes, rate-limits, logs, and accounts for every AI API call and MCP tool invocation across an organization. The solution supports multiple LLM providers including OpenAI, Anthropic, Google Gemini, Azure OpenAI, and AWS Bedrock, with automatic format conversion between different APIs. Its MCP gateway maintains end-user identity when connecting to upstream services, unlike typical proxies that use shared service accounts.
Try asking AI
After installing, here are 7 things you can ask your AI assistant:
When to choose this
Choose ThinkWatch for enterprise environments requiring unified governance across multiple AI providers and MCP tools with comprehensive audit logging and cost tracking.
When NOT to choose this
Not suitable for simple personal use cases or teams that don't require enterprise-grade governance and audit trails.
Tools this server exposes
12 tools extracted from the READMEgithub__create_issueCreate a new GitHub issue with title, body, and labels
notion__create_pageCreate a new page in Notion with title and content
linear__create_ticketCreate a new Linear ticket with title, description, and priority
slack__send_messageSend a message to a Slack channel or direct message
atlassian__create_issueCreate a new Jira issue with summary, description, and priority
cloudflare__list_zonesList Cloudflare DNS zones and their configurations
gitlab__create_merge_requestCreate a new GitLab merge request with title and description
discord__send_messageSend a message to a Discord channel or direct message
google__searchPerform a Google search and return results
postgres__queryExecute SQL queries against a PostgreSQL database
redis__getRetrieve a value from Redis by key
docker__list_containersList Docker containers and their status
Note: Tool names inferred from the MCP Gateway section which lists platforms (GitHub, Notion, Linear, etc.) that MCP tools would connect to. The actual tool names are explicitly mentioned in the format 'platform__action' (e.g., 'github__create_is
Comparable tools
Installation
Installation
ThinkWatch requires Docker and Docker Compose for deployment. Follow these steps:
- Clone the repository:
git clone https://github.com/ThinkWatchProject/ThinkWatch.git
cd ThinkWatch- Start the services:
docker compose up -d- Access the web console at http://localhost:3001
Integration with Claude Desktop
To use ThinkWatch with Claude Desktop, update your Claude Desktop configuration:
{
"mcpServers": {
"thinkwatch": {
"command": "npx",
"args": ["@thinkwatch/mcp-server", "--server", "http://localhost:3000"]
}
}
}Configure your API key in the Claude Desktop settings or set it as an environment variable.
FAQ
- What LLM providers does ThinkWatch support?
- ThinkWatch supports OpenAI, Anthropic, Google Gemini, Azure OpenAI, AWS Bedrock, and any OpenAI-compatible endpoints. It automatically converts between different API formats.
- How does ThinkWatch handle MCP authentication differently from other proxies?
- Unlike typical MCP proxies that use a shared service account token, ThinkWatch maintains per-user OAuth tokens or PATs (Personal Access Tokens) for each upstream service, providing true end-to-end user identity in audit logs.
- Can I use ThinkWatch with existing AI development tools like Cursor or Claude Code?
- Yes, ThinkWorks works as a drop-in replacement for Cursor, Continue, Cline, Claude Code, and OpenAI/Anthropic SDKs by providing a unified proxy interface at a single port.
Compare ThinkWatch with
Last updated · Auto-generated from public README + GitHub signals.