ECONNREFUSED: connection refused — page 2
Nothing was listening at the address Claude Code dialled.
214 issues · 8 open · 206 resolved (96%) · first seen Mar 25, 2025
First response · network
What to do now
Most likely: A DNS, proxy, TLS or socket layer interrupted the connection.
- Check whether the same request works without the current VPN or corporate proxy.
- Verify DNS and certificate trust without disabling TLS verification.
- Capture the hostname and exact socket code before escalating.
These are conservative triage steps, not an official Anthropic fix. Use the issue and workaround evidence below before making a destructive configuration change.
Is this getting better or worse?
This class of problem is converging. 13 new reports in the last 90 days vs 75 in the 90 before — -83%. The open backlog peaked at 43 in 2026-04 and sits at 8 today.
Derived by counting each matching issue from its creation date until it closed. First-party
computation over the anthropics/claude-code tracker — the trend is not published
anywhere upstream.
Which builds report this
188 of these reports name the Claude Code build they were running, spanning 83 releases. Heaviest reporters:
- Claude Code v2.1.114
- Claude Code v2.1.123
- Claude Code v2.1.119
- Claude Code v2.1.92
- Claude Code v2.1.104
- Claude Code v2.1.81
- Claude Code v2.1.29
- Claude Code v2.1.126
Report volume, not a defect count: an issue can mention a build without that build having caused the failure. Versions come from the version string in the bug-report template, matched against real published tags.
Do these get fixed?
Median time to close is 8 days across 206 closures. Of the 206 closures with a recorded reason, 11% were closed as completed and 184 as not-planned or duplicate. The most recent completed fix landed in 2026-10.
Workarounds reported by the community
18 of these issues post a workaround someone says worked. The highest-engagement ones:
- [Bug] 1M context incorrectly requires extra usage on Max plan
- …or charges. Actual Behavior Error is thrown blocking usage. Workaround of running is required, which also incorrectly labels the feature as "Billed as extra usage." Environment - Plan: Max - Model: claude-opus-4-6[1m] -…
Found in the description of #45390 · still open - OAuth token expired immediately after fresh login - cannot use Claude Code (blocking)
- …to succeed, first prompt still returns 401 "token expired" Workaround attempted (unsuccessful): Exit Claude Code rm -rf ~/.claude (clear all credentials) Run claude Run /login Same failure occurs Claude Model Opus Is th…
Found in the description of #19078 · resolved - [Bug] Weekly usage reset time displayed incorrectly for Pro plan users
- …gent: While our Support team is unable to manually reset or work around usage limits, you can learn about best practices here ( If you’ve hit a message limit, you’ll need to wait until the reset time, or you can consider…
Found in the thread of #51222 · resolved - [BUG] Connection-level errors (ECONNRESET, EPIPE, ETIMEDOUT) are never retried — immediate failure with zero recovery
- …+ partial retry reimplementation is the core issue. Partial workaround: PostToolUseFailure hook to auto-retry You can't retry API calls from a hook, but you can detect connection-related failures and alert: Better workar…
Found in the thread of #37077 · resolved - [BUG] http hooks with json don't work.
- …?" — with HTTP hooks failing to reach their endpoint. Their workaround was the same as yours: fall back to command hooks. HookLab works — but it runs on localhost. The HookLab project (a Phoenix LiveView dashboard) recei…
Found in the description of #30613 · resolved
Excerpts are quoted from the linked issue so you can judge relevance before opening it. Threads that only ask for a workaround, or report that there is none, are excluded. These are community suggestions, not official guidance from Anthropic — read the full thread before applying one.
Matching issues 101–200
Ranked by community engagement (reactions weighted over comments).
- [Bug] Anthropic API Error: Extra usage required for Sonnet 4.6 1M context on Max x20 plan
- [Bug] Custom skills not resolved to worktree directory after EnterWorktree
- [Bug] Agents get stuck and stop making progress without utilizing token budget
- [Bug] Rapid token consumption - investigate token usage efficiency
- [Bug] Unexpected human message text appearing in output
- [Bug] Stuck on tasks with "need more time to think" + socket connection/API errors
- [Bug] 1M context model blocked by monthly extra usage limit despite hourly/weekly limits available
- [Bug] Rate limit exceeded on single prompt execution
- [Bug] Repeated permission requests despite previous grants
- [Bug] /clear command incorrectly reports token savings when session usage is low
- [Bug] Claude Code consumes excessive tokens for local filesystem operations available via Google Drive sync
- [Feature Request] Reduce verbosity of output after decision points and queries
- [Bug] Computer-use MCP system warnings leak into TUI rendering on macOS
- [Bug] Excessive credit consumption in CLI (20-100x higher than web version)
- [Bug] Claude Code process hangs and becomes unresponsive requiring terminal restart
- [Bug] 400 Error with image upload breaks session resumption
- [BUG] Claude Code ECONNREFUSED on one Mac — all standard troubleshooting exhausted
- [Feature Request] Use git fixup workflow for amending non-HEAD commits
- [Feature Request] Add support for model selection and prompt injection within conversations
- [Bug] Claude Code agent hallucinating PR review content without verifying actual PR contents
- [Bug] Anthropic API Error: Overloaded
- [Bug] Increased hallucination rate in code generation
- [Bug] Missing command suggestions in terminal after initial message
- [Bug] Anthropic API Error: Invalid cache_control TTL ordering in request
- [Bug] /btw command race condition causes "unknown skill: btw" error
- [Feature Request] Notify users about plugin registry breaking changes in Claude Code
- [Bug] Claude Code CLI fails to initialize or execute
- [Bug] Sub-agent permission_prompt completion not signaled
- [Bug] Claude Code does not display proposed plan/changes before implementation
- [Bug] Model configuration mismatch between CLI setting and session data
- [Bug] Messages fail to send to team-lead after extended use; restart required
- [Bug] Auto-commit runs before user review opportunity
- [Bug] Anthropic API Error: Unsupported image format image/png in base64 encoded content
- [BUG] Claude Code Desktop SSH: Port directive from ~/.ssh/config ignored + unhelpful host verification error
- [Bug] Claude enters infinite loop of redundant file modifications without progress
- [Bug] Context limit exceeded with multiple subagents causes unrecoverable session state
- [Bug] Compact command fails reliably on long conversations, breaks context management workflow
- ECONNRESET kills session — needs auto-retry/reconnect
- [Feature Request] Add /switch-dir command to quickly toggle between added directories
- Bug Report
- Official plugin marketplace (claude-plugins-official) is inaccessible
- [Feature Request] Allow customizable plan file storage location
- [Bug] Feedback submission fails with "Request URL too long" error
- [Bug] Plan mode changes applied to files unexpectedly
- [Bug] Out of Memory crash on macOS 8GB RAM reading CLAUDE.md
- [Bug] Context low warning triggers despite available token budget
- [Bug] Anthropic API Error: Conversation Length Limit Exceeded Unexpectedly
- Bug Report PeriodicExportingMetricReader: metrics export failed
- [BUG] Teleport of a remote session fetches 0 log entries despite the session having a full transcript on claude.ai/code
- [BUG] Claude apps gateway SSRF guard bypasses HTTPS_PROXY by connecting to resolved IP, breaking deployments behind HTTP proxies
- [DOCS] Connection-drops section omits that partial responses are preserved and the spinner no longer gets stuck at "running tool" (v2.1.179)
- [BUG] Agent teams: teammate sessions die with ConnectionRefused (Bun tries IPv6, no IPv4 fallback) while the lead keeps working
- [Feature Request] Improve edge case handling during code debugging
- [Bug] Model generates confident but inaccurate responses on unfamiliar domains without uncertainty signaling
- [Bug] Deleted text from user input appears in Claude Code context
- [Feature Request] Improve code analysis depth for refactoring and duplicate detection across call sites
- [Bug] Model hallucination causing incorrect code generation and failed validation
- [Feature Request] Inline autocorrect for code input
- [Bug] UI Layout Issue: Text and icon overlay not preventable with 91% session limit
- [Bug] Auto-mode disabled by critical permission guardrails, excessive confirmation prompts
- [Bug] Claude Code fails to correlate detected fractures with anatomical structures in radiology analysis
- [Bug] Weekly limit reset does not reset daily limit; 5H rolling window shifts inconsistently
- [Bug] Claude Code: General runtime issues
- [Bug] Context window exhausted mid-session causing unexpected interruption
- [Bug] Remote-control notifications not reaching iPhone and responses not syncing to CLI on Windows
- an[Feature Request] Display full prompt content for approval on mobile devices
- You've hit your limit · resets 7:10am (Australia/Sydney) - while I just barely started work after 1 day break
- [Bug] Model enters infinite loop without progressing on task resolution
- [Bug] Claude 3.5 Sonnet model performance regression compared to previous versions
- [Feature Request] Improve model response quality and guideline adherence
- [Bug] Terminal output exceeds viewport width, requires scrolling to view complete responses
- [Feature Request] Bulk MCP server reconnection for multi-server disconnect scenarios
- [Bug] Claude lacks access to current time and cannot calculate time intervals
- [Bug] Opus 4.7 ships incomplete work with deferred test coverage and missing configuration entries
- [Feature Request] Interactive game conversion workflow for high-performance simulations
- [Bug] Production Stability Regression: Performance Degradation vs Opus 4.5
- [Bug] Claude Code fails to determine current date/time despite timezone context provided
- I appreciate the feedback, but I need actual bug report details to generate a GitHub issue title. Please provide: 1. **Description of the problem** - What went wrong? 2. **Error message** (if applicable) - What was the exact error? 3. **Steps to reproduce
- [Bug] Content filtering incorrectly blocks country names and flags in output
- [Bug] Session stuck in research loop for simple web search tasks
- [Bug] Enter key unresponsive in Google Apps authentication prompt, requires Esc key workaround
- [Bug] Claude executes actions prematurely without awaiting user confirmation
- Claude provided incorrect analysis of job classification performance impact
- [Bug] Excessive API usage with no response completion
- [Bug] Session history repeatedly clearing unexpectedly during active session
- [Bug] Agent mode CPU usage spike at 26% on single-line execution
- [Bug] Context window usage calculation inconsistency across sessions
- Dispatch: messages received but never executed (ECONNREFUSED 127.0.0.1:3845)
- [Bug] Claude Code model regression: degraded code quality and reliability
- [BUG] Interactive mode ignores ANTHROPIC_BASE_URL, connects directly to api.anthropic.com
- [Bug] Agent selects wrong database without user confirmation, wasting API tokens
- Claude Code cannot authenticate from Contabo VPS - Cloudflare blocking claude.ai
- [Bug] Anthropic API Error: Invalid JWT in Edge Function authentication
- [Bug] Notion plugin installation fails with duplicate MCP server configuration error
- [Bug] Supabase environment variables conflicting between staging and production
- [Bug] TUI rendering breaks with usage warning text overflow
- [Bug Report] Unable to process request - missing information
- Bug Report
- [Bug] Anthropic API Error: Missing corresponding tool_use block for tool_result
- Feature Request: Include error details in mcpServerStatus() response