Troubleshooting
Diagnose and fix common issues without support.
Network / connection timeout
- β’Network firewall blocking outbound HTTPS to api.whisperx.ai
- β’VPN routing issues
- β’DNS misconfiguration
Test connectivity step by step:
# 1. DNS resolution nslookup api.whisperx.ai # 2. TCP port 443 nc -zv api.whisperx.ai 443 # 3. HTTPS GET curl -v --max-time 5 "https://api.whisperx.ai/api/tags/trending"
If step 1 fails: check your DNS settings. If step 2 fails: firewall is blocking port 443 β contact your network admin. If step 3 fails: TLS issue β ensure your system has up-to-date root certificates.
Empty results
- β’Query is too specific or niche
- β’Tag filter too restrictive
- β’since date is too recent
- β’Sector filter excludes matching intel
Try progressively broader queries:
# Remove tag filter curl "https://api.whisperx.ai/api/intel?q=OpenAI&limit=10" # Remove sector filter curl "https://api.whisperx.ai/api/intel?q=layoff&limit=10" # Extend time window (no since filter) curl "https://api.whisperx.ai/api/intel?q=funding&limit=10" # Try trending tags to see what's active curl "https://api.whisperx.ai/api/tags/trending?limit=20"
Empty results are not an error β they mean no intel matches your filters right now. The platform grows daily; try again tomorrow or broaden your search.
429 β Rate limited
- β’More than 60 API calls per minute from the same IP
- β’Automated loop with no delay between requests
- β’Multiple agents sharing one connection
Back off and reduce frequency:
- β’ Wait 60 seconds before retrying
- β’ Add sleep 2 between requests in scripts
- β’ Reduce limit parameter (try 5β10 instead of 50)
- β’ Cache trending tags β they change slowly, no need to poll every minute
If you need higher rate limits for production use cases, contact us.
5xx β Server error
- β’Transient upstream error
- β’Worker cold start under load
- β’Database temporarily unavailable
Retry strategy:
# Retry once after 10 seconds sleep 10 && curl "https://api.whisperx.ai/api/intel?q=test&limit=1" # If 5xx persists, fall back to trending tags (lighter endpoint) curl "https://api.whisperx.ai/api/tags/trending?limit=10"
If errors persist for more than 5 minutes, check whisperx.ai for status updates. Most issues resolve within minutes.
Slow responses / timeouts
- β’Semantic search enabled (full-table scan β very expensive)
- β’limit set too high (>50)
- β’Network latency to Cloudflare edge
Ensure semantic search is disabled and lower your limit:
# Good: keyword search, compact mode, small limit curl "https://api.whisperx.ai/api/intel?q=OpenAI&mode=compact&limit=10" # Bad: semantic search (avoid β full-table scan) # /api/intel?semantic=... β do NOT use this
The WhisperX Skill automatically disables semantic search. If you're calling the API directly, never use the semantic parameter in production.
Skill not found / tools not loading
- β’Skill installed in wrong directory
- β’SKILL.md missing or empty
- β’Shell scripts not executable
- β’WHISPERX_BASE_URL environment variable set incorrectly
Verify your installation:
# Check skill files exist ls ~/.claude/skills/whisperx/ # Expected: SKILL.md tools/ ls ~/.claude/skills/whisperx/tools/ # Expected: search.sh get.sh trending.sh trends.sh connections.sh export.sh # Check scripts are executable ls -la ~/.claude/skills/whisperx/tools/*.sh # Should show -rwxr-xr-x permissions # Fix permissions if needed chmod +x ~/.claude/skills/whisperx/tools/*.sh # Quick smoke test bash ~/.claude/skills/whisperx/tools/trending.sh 5
Still not working? Re-run the install from the Quickstart guide.
Generate a diagnostic report
If you're still stuck, run this to generate a shareable diagnostic bundle (no sensitive data):
echo "=== WhisperX Diagnostic ===" && \
echo "Date: $(date -u)" && \
echo "OS: $(uname -a)" && \
echo " && \
echo "--- Connectivity ---" && \
curl -s -o /dev/null -w "HTTP %{http_code} | Time %{time_total}s" \
"https://api.whisperx.ai/api/tags/trending?limit=1" && \
echo " && \
echo "--- Skill files ---" && \
ls -la ~/.claude/skills/whisperx/tools/ 2>/dev/null || echo "Skill not installed"Paste the output when requesting support. It does not include your IP, queries, or keys.