[BUG] WebFetch tool unable to access any external documentation sites

[Muelmue]

Preflight Checklist

• I have searched existing issues and this hasn't been reported yet
• This is a single bug report (please file separate reports for different bugs)
• I am using the latest version of Claude Code

What's Wrong?

When attempting to use the WebFetch tool to access external documentation sites, Claude Code consistently returns errors indicating it cannot fetch from any external domains. This severely limits the tool's ability to provide accurate, up-to-date information from official documentation sources.

The command fails immediately with the error message: Claude Code is unable to fetch from [domain] for all tested sites.

What Should Happen?

Claude Code should successfully fetch content from common documentation sites and technical resources needed for software development tasks, particularly:

• Official Claude Code documentation (docs.claude.com)
• Programming language documentation (docs.oracle.com, developer.mozilla.org)
• Code repositories (github.com)

Error Messages/Logs

Steps to Reproduce

1. Start a Claude Code session
2. Request information that requires fetching external documentation:
   User: "Can you check the Claude Code documentation about model settings?"
3. Claude Code attempts to use WebFetch tool:
   WebFetch(url="https://docs.claude.com/en/docs/claude-code/configuration")
4. Error appears: "Claude Code is unable to fetch from docs.claude.com"

Note: This issue occurs with ALL external websites, not just Claude's documentation.

Error Messages/Logs

<function_results>
Claude Code is unable to fetch from docs.claude.com
</function_results>

<function_results>
Claude Code is unable to fetch from docs.oracle.com
</function_results>

<function_results>
Claude Code is unable to fetch from developer.mozilla.org
</function_results>

<function_results>
Claude Code is unable to fetch from github.com
</function_results>

Claude Model

Opus

Is this a regression?

I don't know

Last Working Version

No response

Claude Code Version

1.0.127 (Claude Code)

Platform

Anthropic API

Operating System

Windows

Terminal/Shell

IntelliJ IDEA terminal

Additional Information

model: Claude Opus 4.1 (claude-opus-4-1-20250805)

[github-actions]

Found 2 possible duplicate issues:

1. [BUG] WebFetch "Unable to verify domain is safe" error #6388
2. web-fetch not working in corporate environment #6188

This issue will be automatically closed as a duplicate in 3 days.

• If your issue is a duplicate, please close it and 👍 the existing issue instead
• To prevent auto-closure, add a comment or 👎 this comment

🤖 Generated with Claude Code

[map-boop]

I also just started experiencing this issue on Mac today

[jwesleye]

Had the same problem on MacOS. Attempted 30 plus different URLs with 100% failing to be able to access the content from Claude code. All of them worked from browser or Claude.ai app. The fails were unique to Claude code.

[econwatch]

This is also a failure on my side.
This request to read the full post went round and round until we determined that Claude is incapable of reading full web content due to its limitations - this vastly reduces Claude's internet search ability skipping vital information.
Was looking for the correct format for config.yml for Frigate NVR configuration. Took 2 hours longer than necessary with Claude going off on tangents. Again, I should have just RTFM and github error reports. Claude is incapable of this consistently.

Re: URL - blakeblackshear/frigate#18755

So what have we learned about your ability to retrieve information from the internet as discussed re: your inability to read the full text of a post?[WHAT WE LEARNED ABOUT WEB_FETCH LIMITATIONS]

JavaScript-rendered content is invisible

GitHub discussions load comments dynamically via JavaScript
web_fetch only captures initial HTML, not dynamically loaded content
The ?plain=1 parameter did not help

Web_fetch cannot see what humans see

You could see hawkeye217's full comment in your browser
I could only see "From the docs:" with everything after truncated
The critical configuration code was completely missing from my fetch

This is a fundamental limitation, not a configuration issue

No parameter adjustment would have worked
GitHub's architecture specifically prevents this
Many modern websites use similar JavaScript-heavy designs

The practical solution

Direct paste from user is fastest and most reliable
Trying multiple fetch approaches wasted significant time
I should have recognized this limitation earlier and asked for direct paste immediately

[BROADER LESSON]
I wasted your time trying to fetch content that was fundamentally inaccessible to my tools, then made multiple contradictory claims based on incomplete information.
When web_fetch fails to retrieve complete content, I should immediately request direct paste rather than continuing to make claims based on partial data.

[wangyum]

Why Claude Code is unable to fetch from github.com? could you fix it?

⏺ I apologize for the confusion. The issue is that the WebFetch tool has a restriction that prevents it from fetching content from github.com. This is a built-in limitation of the tool itself, not something I can fix through code
changes.

However, I notice from my tool documentation that there's an important note:

"IMPORTANT: If an MCP-provided web fetch tool is available, prefer using that tool instead of this one, as it may have fewer restrictions. All MCP-provided tools start with 'mcp__'."

Unfortunately, I don't see any MCP-provided tools available in my current environment that could help fetch from GitHub.

Alternative approaches I can suggest:

1. If you have gh CLI installed: I tried to use it but got "command not found". If you install GitHub CLI (brew install gh on macOS), I can use it to fetch PR details.
2. You could copy-paste the PR content: If you can access the PR and copy the relevant parts (description, key comments, code changes), I can analyze that.
3. You could download the PR as a patch: If you have the PR patch file locally, I can read and analyze it.

The restriction on fetching from github.com appears to be a security/policy decision in the WebFetch tool design, likely to prevent scraping or automated access to GitHub. This isn't something I can work around within the current
tool limitations.

gh CLI works.

[github-actions]

This issue has been inactive for 30 days. If the issue is still occurring, please comment to let us know. Otherwise, this issue will be automatically closed in 30 days for housekeeping purposes.

[github-actions]

This issue has been automatically closed due to 60 days of inactivity. If you're still experiencing this issue, please open a new issue with updated information.