[BUG] API Error
Preflight Checklist
- [x] I have searched existing issues and this hasn't been reported yet
- [x] This is a single bug report (please file separate reports for different bugs)
- [x] I am using the latest version of Claude Code
What's Wrong?
Bug Description
Claude Desktop app suddenly started showing "API Error" when attempting to use Cowork/agent features. This worked perfectly before a recent app update and stopped working without any changes to my network configuration.
What's Wrong?
After a recent Claude Desktop update, the app fails to connect to the API with various errors, despite the network connection to Anthropic's servers being fully functional (verified via terminal).
What Should Happen?
[BUG] Claude Desktop App - API Error after recent update (network connectivity verified working)
Preflight Checklist
- [x] I have searched existing issues and this hasn't been reported yet
- [x] This is a single bug report
- [x] I am using the latest version of Claude Desktop
Environment
- Claude Desktop version: Latest (as of January 19, 2026)
- Operating System: macOS
- Location: Mexico (Totalplay ISP)
- VPN/Proxy: Tested both with and without Cloudflare WARP - same issue
Bug Description
Claude Desktop app suddenly started showing "API Error" when attempting to use Cowork/agent features. This worked perfectly before a recent app update and stopped working without any changes to my network configuration.
What's Wrong?
After a recent Claude Desktop update, the app fails to connect to the API with various errors, despite the network connection to Anthropic's servers being fully functional (verified via terminal).
Error Messages from Logs (~/Library/Logs/Claude/claude.ai-web.log):
2026-01-19 12:53:33 [error] [REACT_QUERY_CLIENT] QueryClient error: RegistryFetchError: Registry fetch failed: TypeError: Failed to fetch (api.anthropic.com)
2026-01-19 12:53:33 [error] [MCP] Registry fetch failed with diagnostics: [object Object]
2026-01-19 12:59:12 [error] [REACT_QUERY_CLIENT] QueryClient error: TypeError: Failed to fetch (claude.ai)
Additionally, there are CSP (Content Security Policy) errors that appear to be an internal bug:
2026-01-19 12:48:07 [error] Refused to connect to 'https://a-api.anthropic.com/v1/m' because it violates the following Content Security Policy directive: "connect-src 'self' https://api.segment.io https://*.segment.io https://*.segment.com https://a-cdn.anthropic.com https://*.google.com https://*.facebook.com https://*.facebook.net https://*.doubleclick.net".
2026-01-19 12:48:07 [error] Fetch API cannot load https://a-api.anthropic.com/v1/m. Refused to connect because it violates the document's Content Security Policy.
Note: The CSP directive doesn't include a-api.anthropic.com in the allowed list, which seems like a configuration error in the app itself.
Network Diagnostics (Proving it's NOT a network issue)
I ran extensive diagnostics from terminal. All tests PASSED:
1. DNS Resolution ✅
api.anthropic.com → 160.79.104.10
console.anthropic.com → 160.79.104.10
claude.ai → 160.79.104.10
statsig.anthropic.com → 34.36.57.103
2. TCP Connectivity (Port 443) ✅
api.anthropic.com: Connection to api.anthropic.com port 443 [tcp/https] succeeded!
console.anthropic.com: Connection to console.anthropic.com port 443 [tcp/https] succeeded!
claude.ai: Connection to claude.ai port 443 [tcp/https] succeeded!
statsig.anthropic.com: Connection to statsig.anthropic.com port 443 [tcp/https] succeeded!
3. TLS 1.3 Handshake ✅
Connecting to 160.79.104.10
depth=2 C=US, O=Google Trust Services LLC, CN=GTS Root R4
verify return:1
depth=1 C=US, O=Google Trust Services, CN=WE1
verify return:1
depth=0 CN=api.anthropic.com
verify return:1
CONNECTED(00000005)
4. HTTP API Response ✅
curl to api.anthropic.com/v1/messages returns HTTP 405 (Method Not Allowed) - expected without POST
curl to a-api.anthropic.com returns HTTP 404 - server responds
curl to a-cdn.anthropic.com returns HTTP 404 - server responds
5. Traceroute ✅
traceroute to api.anthropic.com (160.79.104.10), 15 hops max
1 192.168.100.1 - 0.958 ms
2 10.249.0.1 - 2.298 ms
...
7 160.79.104.10 - 6.464 ms ← Reaches Anthropic in ~6ms
Steps to Reproduce
- Update Claude Desktop to the latest version
- Open Claude Desktop
- Try to use Cowork or any agent feature
- Observe "API Error" message
Expected Behavior
Claude Desktop should connect to the API successfully, as it did before the recent update and as terminal connections prove is possible.
Actual Behavior
The app shows "API Error" despite:
- All network connectivity tests passing from terminal
- The same network working perfectly for claude.ai in browser
- No firewall or proxy blocking (tested with/without WARP)
Troubleshooting Already Attempted
- [x] Cleared app cache (
~/Library/Caches/Claude/and~/Library/Application Support/Claude/Cache/) - [x] Signed out and signed back in
- [x] Tested with Cloudflare WARP connected
- [x] Tested without WARP
- [x] Verified no ANTHROPIC_API_KEY environment variable conflicts
- [x] Verified DNS resolution works (using Cloudflare 1.1.1.1)
- [x] Reinstalled the app
Key Observations
- This is NOT a network issue - All connectivity tests pass from terminal
- The CSP error is an internal bug -
a-api.anthropic.comis not in the CSP allowlist - This worked before - The exact same network configuration worked until the recent update
- Browser works fine - claude.ai in Chrome works perfectly on the same network
Hypothesis
The recent app update introduced either:
- A CSP misconfiguration that blocks
a-api.anthropic.com - A WebView/Electron networking issue that doesn't affect terminal requests
- An OAuth token handling bug (similar to issue #18444)
System Information
macOS with latest Claude Desktop
ISP: Totalplay (Mexico)
Router: Standard config, no port blocking
DNS: Cloudflare 1.1.1.1 (verified working)
No corporate firewall/proxy
Logs
Full logs available at ~/Library/Logs/Claude/claude.ai-web.log
The CSP errors have been occurring since at least January 14, 2026, based on log timestamps.
Error Messages/Logs
What's Wrong?
After a recent Claude Desktop update, the app fails to connect to the API with various errors, despite the network connection to Anthropic's servers being fully functional (verified via terminal).
Error Messages from Logs (~/Library/Logs/Claude/claude.ai-web.log):
2026-01-19 12:53:33 [error] [REACT_QUERY_CLIENT] QueryClient error: RegistryFetchError: Registry fetch failed: TypeError: Failed to fetch (api.anthropic.com)
2026-01-19 12:53:33 [error] [MCP] Registry fetch failed with diagnostics: [object Object]
2026-01-19 12:59:12 [error] [REACT_QUERY_CLIENT] QueryClient error: TypeError: Failed to fetch (claude.ai)
Additionally, there are CSP (Content Security Policy) errors that appear to be an internal bug:
2026-01-19 12:48:07 [error] Refused to connect to 'https://a-api.anthropic.com/v1/m' because it violates the following Content Security Policy directive: "connect-src 'self' https://api.segment.io https://*.segment.io https://*.segment.com https://a-cdn.anthropic.com https://*.google.com https://*.facebook.com https://*.facebook.net https://*.doubleclick.net".
2026-01-19 12:48:07 [error] Fetch API cannot load https://a-api.anthropic.com/v1/m. Refused to connect because it violates the document's Content Security Policy.
Note: The CSP directive doesn't include a-api.anthropic.com in the allowed list, which seems like a configuration error in the app itself.
Network Diagnostics (Proving it's NOT a network issue)
I ran extensive diagnostics from terminal. All tests PASSED:
1. DNS Resolution ✅
api.anthropic.com → 160.79.104.10
console.anthropic.com → 160.79.104.10
claude.ai → 160.79.104.10
statsig.anthropic.com → 34.36.57.103
2. TCP Connectivity (Port 443) ✅
api.anthropic.com: Connection to api.anthropic.com port 443 [tcp/https] succeeded!
console.anthropic.com: Connection to console.anthropic.com port 443 [tcp/https] succeeded!
claude.ai: Connection to claude.ai port 443 [tcp/https] succeeded!
statsig.anthropic.com: Connection to statsig.anthropic.com port 443 [tcp/https] succeeded!
3. TLS 1.3 Handshake ✅
Connecting to 160.79.104.10
depth=2 C=US, O=Google Trust Services LLC, CN=GTS Root R4
verify return:1
depth=1 C=US, O=Google Trust Services, CN=WE1
verify return:1
depth=0 CN=api.anthropic.com
verify return:1
CONNECTED(00000005)
4. HTTP API Response ✅
curl to api.anthropic.com/v1/messages returns HTTP 405 (Method Not Allowed) - expected without POST
curl to a-api.anthropic.com returns HTTP 404 - server responds
curl to a-cdn.anthropic.com returns HTTP 404 - server responds
5. Traceroute ✅
traceroute to api.anthropic.com (160.79.104.10), 15 hops max
1 192.168.100.1 - 0.958 ms
2 10.249.0.1 - 2.298 ms
...
7 160.79.104.10 - 6.464 ms ← Reaches Anthropic in ~6ms
Steps to Reproduce
Steps to Reproduce
Update Claude Desktop to the latest version
Open Claude Desktop
Try to use Cowork or any agent feature
Observe "API Error" message
Claude Model
Not sure / Multiple models
Is this a regression?
Yes, this worked in a previous version
Last Working Version
1.1.381 (previous to this one)
Claude Code Version
1.1.381
Platform
Anthropic API
Operating System
macOS
Terminal/Shell
Terminal.app (macOS)
Additional Information
_No response_
This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗