VS Code Context7 MCP: "TypeError: fetch failed" to https://mcp.context7.com/mcp
In the Context7 MCP plugin for VS Code I get the error: "Connection state: Error Error sending message to https://mcp.context7.com/mcp (opens external site): TypeError: fetch failed". Another MCP client shows "Failed to reconnect to context7". Is this a server outage, a config problem, or a proxy issue, and was it resolved?
Originally asked on GitHub: https://github.com/upstash/context7/issues/803
6 answers
Status: explained for the original report. On 2025-10-20 a Context7 maintainer said the service was down because of an AWS outage and asked the user to retry. However, the thread does not confirm a retry fixed it: a user on 2025-10-22 still could not connect via MCP, and another on 2025-10-24 reported the same "fetch failed" error. The issue was later closed as stale in January 2026.
If you see this error, first retry after a short wait. The thread's only confirmed external cause is the AWS outage. For persistent cases, one user said they could connect from another machine and guessed their company proxy (which modifies certificates) was interfering; that was not confirmed. A maintainer then said Context7 supports proxy environment variables and pointed to the Context7 README, but the thread does not name the variable names or show a working configuration.
General advice: to tell a server outage from a local network/proxy issue, test https://mcp.context7.com/mcp (opens external site) from another machine or network. If it works elsewhere, the problem is likely on your side (proxy, certificates, firewall); if it fails everywhere, the service may still be having trouble.
Did this work for you?What @skiabox tried on GitHub (Didn't work)
The reporter showed the VS Code Context7 plugin error "Connection state: Error Error sending message to https://mcp.context7.com/mcp (opens external site): TypeError: fetch failed" and opened the issue.
Source: https://github.com/upstash/context7/issues/803 (opens external site)
Did this work for you?What @enesgules tried on GitHub (Outcome unclear)
A maintainer said the service was down on 2025-10-20 due to an AWS outage and asked the user to try again. No one in the thread confirmed that a retry fixed it.
Source: https://github.com/upstash/context7/issues/803#issuecomment-3422243685 (opens external site)
Did this work for you?What @lfilipiuk tried on GitHub (Didn't work)
A user reported they still could not connect via MCP and saw "Failed to reconnect to context7", so the outage explanation did not cover their situation.
Source: https://github.com/upstash/context7/issues/803#issuecomment-3434056723 (opens external site)
Did this work for you?What @dermoritz tried on GitHub (Partly worked)
Another user with the same error said they could connect from a different machine and suspected their company proxy was interfering with certificates. The proxy cause was never confirmed.
Source: https://github.com/upstash/context7/issues/803#issuecomment-3441270442 (opens external site)
Did this work for you?What @enesgules tried on GitHub (Outcome unclear)
A maintainer said Context7 supports proxy environment variables and referred the user to the README, but the thread does not list the variable names or a working config.
Source: https://github.com/upstash/context7/issues/803#issuecomment-3441578735 (opens external site)
Did this work for you?
Related questions
- How does context7 decide which docs are trusted? What is trustScore?
- Cursor MCP "Failed to create client" with @upstash/context7-mcp using npx: fix
- Context7 MCP server rate limits my first request in Trae — is it my config or Context7?
- Does Context7 support GitLab repositories now, and where do I add one?
- Is there an official public Docker image for the context7 MCP server?