Common Issues

Auth Failures

Expired session token

Symptoms: the chat panel shows "Session expired" or "Unauthorized". API requests return HTTP 401.

• Sign out and sign back in from the chat panel (click your avatar > Sign Out).
• If the problem persists, clear the stored credentials: open the Command Palette (Cmd/Ctrl + Shift + P) and run "Creor: Clear Credentials".
• Check that your system clock is accurate. Token validation fails if your clock is more than 5 minutes off.

Invalid API key

Symptoms: API requests return HTTP 401 with "Invalid API key" in the response body.

• Verify the key has not been revoked in Dashboard > Settings > API Keys.
• Check that you are using the correct key for the correct workspace.
• Ensure the key prefix matches the environment: cr_live_ for production, cr_test_ for staging.
• If using environment variables, verify the variable is set correctly (echo $CREOR_API_KEY).

BYOK provider key rejected

Symptoms: "Authentication failed" when using your own API key for a provider like Anthropic or OpenAI.

• Verify the key is correct and has not expired or been rotated.
• Check that the key has the required permissions (e.g., OpenAI keys need API access, not just ChatGPT Plus).
• For organization-scoped keys, verify the organization ID is set correctly.
• Try the key directly with the provider's API to confirm it works outside Creor.

Model Errors

Rate limit exceeded

Symptoms: "Rate limit exceeded" or HTTP 429. Requests fail temporarily.

• Wait 30-60 seconds and try again. Rate limits reset automatically.
• If using BYOK, check your provider's rate limit tier. Higher tiers have higher limits.
• Switch to a different model temporarily. Rate limits are per-model.
• If using the Creor Gateway, upgrade your plan for higher rate limits.

Model not available

Symptoms: "Model not found" or the model does not appear in the picker.

• Check that the model is available on your plan. Free plan users have access to a subset of models.
• For BYOK providers, verify the model name matches exactly (e.g., "claude-sonnet-4-20250514" not "claude-sonnet").
• Some models require specific provider access (e.g., Claude requires an Anthropic API key or Creor Gateway).
• The model may have been deprecated. Check the provider's documentation for the current model list.

Context window exceeded

Symptoms: "Context length exceeded" or the agent stops mid-response.

• Start a new session. Long conversations accumulate context that can exceed limits.
• Use a model with a larger context window (e.g., Claude Sonnet supports up to 200K tokens).
• Scope your requests more narrowly to reduce the amount of code the agent reads.
• Enable session compaction in settings to automatically summarize older turns.

Credit Exhaustion

When your credit balance reaches zero, gateway requests and cloud agent runs are paused.

How to resume

• Add credits: Dashboard > Billing > Add Credits. Credits are available within seconds.
• Wait for your billing cycle: subscription credits are added at the start of each cycle.
• Switch to BYOK: configure your own API keys in Settings > Providers to bypass gateway billing.

Preventing credit exhaustion

• Set up low-balance alerts: Dashboard > Billing > Alerts.
• Set spending limits per member, per API key, or per month.
• Monitor the Usage page to understand your consumption patterns.
• Use cheaper models (Haiku, GPT-4o mini) for simple tasks.
• Enable auto-purchase: Dashboard > Billing > Auto-Purchase to automatically buy credits when your balance is low.

Connection Issues

Engine not connecting

Symptoms: the chat panel shows "Connecting to engine..." indefinitely or "Engine not available".

• Check the Output panel (View > Output > Creor Engine) for error messages.
• The engine may have crashed. Restart it: Command Palette > "Creor: Restart Engine".
• Check that port 4096 (default engine port) is not in use by another process.
• On macOS, verify the engine binary has execution permissions.

Server unreachable (gateway)

Symptoms: "Server unreachable" or "Network error" when sending messages via the Creor Gateway.

• Check your internet connection.
• Verify that api.creor.ai is reachable: try opening creor.ai in a browser.
• If you are behind a corporate proxy or firewall, see the Network & Proxy troubleshooting page.
• Check the Creor status page (status.creor.ai) for ongoing incidents.

Engine Startup

Engine binary not found

Symptoms: "Failed to start engine: ENOENT" in the Output panel.

• The engine binary may not have been extracted correctly during installation. Reinstall Creor.
• On macOS, the binary may be quarantined. Run: xattr -dr com.apple.quarantine /Applications/Creor.app
• On Linux, verify the binary has execute permission: chmod +x /path/to/opencode

Engine crashes on startup

Symptoms: the engine starts and immediately exits. The Output panel shows an error or stack trace.

• Check if Bun is installed and accessible (the engine requires Bun 1.3.8+).
• Look for port conflicts on port 4096.
• Check disk space -- the engine needs at least 100 MB free for temporary files.
• On Apple Silicon Macs, verify you downloaded the arm64 build, not the x64 build.

Performance Issues

Slow responses

If the agent takes a long time to respond, the bottleneck is usually the LLM provider, not Creor itself.

• Switch to a faster model. Haiku and GPT-4o mini respond in 1-3 seconds for most queries.
• Check the provider's status page for degraded performance.
• If using BYOK, your rate limit tier may throttle long-running requests.
• Reduce context size by starting a new session or scoping requests more narrowly.

High memory usage

Creor runs multiple processes (editor, engine, extensions). High memory usage is normal for large workspaces.

• Close unused editor tabs. Each open file consumes memory for syntax highlighting and IntelliSense.
• Disable extensions you are not actively using.
• If codebase indexing (RAG) is running, memory will be higher temporarily. It drops after indexing completes.
• Restart Creor to reclaim memory from leaked processes or extensions.

Getting Help

If the solutions above do not resolve your issue, here is how to get additional help.

• Discord: Join the Creor Discord server for community support and real-time help.
• GitHub Issues: Report bugs at github.com/modhisathvik7733/creor-app/issues.
• Email support: support@creor.ai (Starter and Pro plans get priority response).
• Status page: status.creor.ai for real-time service health.

When Reporting a Bug

Include this information to help us diagnose the issue faster:

• Creor version (Help > About).
• Operating system and version.
• Steps to reproduce the issue.
• Output panel logs (View > Output > Creor Engine).
• Screenshot or screen recording if applicable.