# Claude Settings Audit Guide > **Purpose**: Condensed reference for Claude.ai Projects > **Source**: 00-claude-settings-audit skill > **Version**: 1.0 Comprehensive reference for diagnosing Claude Desktop errors and optimizing context usage. ## Error Quick Reference | Error Type | Message Pattern | Category | |------------|-----------------|----------| | Context too large | "Exceed response limit" | Input | | Output interrupted | "Response could not be fully generated" | Output | | Length exceeded | "Message will exceed the length limit" | Input | | Conversation limit | "Conversation reached maximum length" | Input | | File too large | "Files larger than 10mb" | File | | Rate limited | "5-hour limit reached" | Usage | | Server overload | "Unexpected capacity constraints" | Server | | MCP failure | "Error connecting to [ServerName]" | MCP | | Account issue | "Account has been disabled" | Account | ## Quick Diagnosis Tree ``` Error occurred? │ ├─ BEFORE response started? │ ├─ "Exceed response limit" → Reduce input context │ ├─ "Length limit" → Break into smaller messages │ ├─ "File too large" → Reduce file size or excerpt │ ├─ "Capacity constraints" → Wait and retry │ └─ "MCP error" → Check MCP configuration │ ├─ DURING response? │ ├─ "Could not be fully generated" → Request smaller output │ └─ Output suddenly stopped → Retry or chunk request │ └─ BEFORE even sending? ├─ "5-hour limit reached" → Wait for reset ├─ "Messages left" → Near usage limit └─ "Account disabled" → Contact support ``` --- ## Token Limits by Model | Model | Context Window | Output Limit | Practical Available | |-------|---------------|--------------|---------------------| | Claude 3.5 Sonnet | 200K tokens | ~8K tokens | ~150K after system | | Claude 3 Opus | 200K tokens | ~4K tokens | ~150K after system | | Claude 3 Haiku | 200K tokens | ~4K tokens | ~160K after system | --- # ERROR CATEGORIES ## 1. Input/Context Errors ### "Exceed response limit" **Causes:** - Loaded context too large (MCP tools + skills + files) - Conversation history accumulated - Multiple large files attached **Solutions:** 1. Reduce MCP servers (max 5 recommended) 2. Add `serverInstructions` to all MCP servers 3. Start fresh conversation 4. Attach file excerpts instead of full files ### "Message will exceed the length limit" **Solutions:** 1. Break content into smaller chunks 2. Summarize key sections before sending 3. Start a new conversation ### "Conversation reached maximum length" **Solutions:** 1. Start a new conversation 2. Copy essential context to new chat --- ## 2. Output Errors ### "Response could not be fully generated" **Causes:** - Output exceeded ~4K-8K token limit - Safety filter triggered - Network interruption - Server timeout **Solutions:** 1. Request smaller chunks of output 2. Add constraints: "In under 500 words..." 3. Be more specific in scope 4. Retry once before rephrasing **Prompting techniques:** - "Provide a concise summary..." - "List the top 10..." - "First, outline the structure" - "Use bullet points, not paragraphs" --- ## 3. File Errors ### "Files larger than 10MB" **Solutions:** 1. Compress file if possible 2. Split into smaller files 3. Extract text content only ### "Message is X% over the length limit" **Solutions:** 1. Extract relevant excerpts only 2. Summarize large documents 3. Split across multiple messages --- ## 4. Usage Limit Errors ### "5-hour limit reached" **Solutions:** 1. Wait until reset time shown 2. Upgrade plan for higher limits 3. Optimize prompts to use fewer tokens ### "X messages left until [time]" **Solutions:** 1. Combine multiple questions into one 2. Wait for reset 3. Consider paid plan --- ## 5. Server/Capacity Errors ### "Unexpected capacity constraints" **Solutions:** 1. Wait 2-5 minutes and retry 2. Try during off-peak hours 3. Check status.claude.com for incidents ### HTTP Error Codes | Code | Meaning | Solution | |------|---------|----------| | 500 | Internal Server Error | Wait and retry | | 429 | Too Many Requests | Slow down, wait | | 403 | Forbidden | Check account status | | 400 | Bad Request | Check input format | --- ## 6. MCP Connection Errors ### "Error connecting to [ServerName]" **Causes:** 1. Server not running 2. Incorrect configuration 3. Transport type mismatch (stdio vs HTTP) 4. Authentication failure **Diagnosis steps:** 1. Check server logs: - macOS: `~/Library/Logs/Claude/mcp-server-SERVERNAME.log` - Windows: `%APPDATA%\Claude\logs\mcp-server-SERVERNAME.log` 2. Verify configuration in `claude_desktop_config.json` 3. Test server independently **Common fixes:** | Issue | Fix | |-------|-----| | Transport mismatch | Add `"transport": "http"` to config | | Path with spaces | Use escaped paths | | Missing env vars | Add to `"env"` in config | | Server not installed | Run `npm install -g` first | **Critical:** Restart Claude Desktop after config changes. --- ## 7. Account Errors ### "Account has been disabled" **Solutions:** 1. Contact Anthropic support 2. Review Terms of Service 3. Do not create new accounts ### "Error logging you in" **Solutions:** 1. Disable VPN 2. Clear browser cache and cookies 3. Try incognito window 4. Check status.claude.com --- ## 8. Output Quality Issues ### Output Suddenly Stops **Solutions:** 1. Ask "Please continue from where you stopped" 2. Request smaller chunks upfront 3. Be more specific in scope ### Gibberish Output **Solutions:** 1. Simplify prompt 2. Start new conversation 3. Rephrase request differently ### Claude Refuses to Respond **Solutions:** 1. Rephrase more clearly 2. Explain legitimate use case 3. Break into smaller requests --- # TOKEN BUDGET GUIDELINES ## MCP Server Token Estimates | Server | Tokens (approx) | |--------|-----------------| | Playwright | ~13,500 | | Notion | ~5,000 | | GitHub | ~18,000 | | Filesystem | ~4,000 | | Zapier | ~25,000+ (avoid) | ## serverInstructions Pattern Every MCP server should have: ```json "serverInstructions": "[Purpose]. Use for: [case1], [case2]. Keywords: [kw1], [kw2]" ``` Example: ```json "serverInstructions": "Browser automation. Use for: screenshots, page testing. Keywords: browser, DOM, screenshot" ``` ## Budget Calculator ``` System overhead: ~10,000 tokens (fixed) MCP servers: [count] × ~3,000 = [estimate] Loaded skills: [count] × ~500 = [estimate] Project instructions: ~[estimate] ──────────────────────────────────────────── Baseline total: [sum] Available for work: 200,000 - [sum] = [remaining] ``` **Targets:** - Baseline: < 40,000 tokens (20%) - Available: > 160,000 tokens (80%) --- # PREVENTION CHECKLIST ## Context/Input - [ ] Maximum 5 MCP servers active - [ ] All servers have serverInstructions - [ ] Skills under 1,000 words - [ ] Start fresh conversations for new topics - [ ] Files under 10MB, prefer excerpts ## Output - [ ] Break large requests into chunks - [ ] Use specific scope in prompts - [ ] Paginate lists (50 items max) - [ ] Retry once before rephrasing ## Usage Limits - [ ] Monitor "messages left" warnings - [ ] Combine related questions - [ ] Use concise prompts ## MCP Stability - [ ] Test servers before adding - [ ] Restart app after config changes - [ ] Keep servers updated ## General - [ ] Check status.claude.com when issues occur - [ ] Keep Claude Desktop updated - [ ] Back up important conversations --- # CONFIGURATION LOCATIONS ## Config Files **macOS:** ``` ~/Library/Application Support/Claude/claude_desktop_config.json ``` **Windows:** ``` %APPDATA%\Claude\claude_desktop_config.json ``` ## Log Files **macOS:** ``` ~/Library/Logs/Claude/mcp-server-SERVERNAME.log ``` **Windows:** ``` %APPDATA%\Claude\logs\mcp-server-SERVERNAME.log ``` --- *Reference: Claude Settings Audit Skill v1.0* *Last updated: 2025-02*