Troubleshooting

Common issues and solutions when using Dino.

Provider & Model Errors

“API key is required” / “Authentication token is required”

Your API key is missing or invalid.

1. Open Settings → Models & Providers
2. Select the provider and re-enter your API key
3. Try sending a message — if it works, the key is valid

HTTP 429 (Rate Limit)

Dino automatically retries 429s up to 3 times with exponential backoff. If you still see this:

• Wait a few minutes — most rate limits reset quickly
• Check your provider’s usage dashboard for quota status
• Coding plans may throttle at daily caps — switch to an alternate provider temporarily
• API key providers: check if you’ve hit your plan’s rate limit tier

HTTP 400 (Bad Request)

The request format doesn’t match what the model expects. Common causes:

• Wrong model name — the model was deprecated or renamed (e.g., gpt-4 → gpt-4.1)
• Wrong SDK type — using google SDK type with an OpenAI-compatible endpoint, or vice versa. Check the SDK type in your provider settings matches the endpoint
• Model doesn’t support tools — some smaller or older models don’t support function calling. Dino’s tool-based workflow requires a model that supports it

HTTP 401 / 403 (Unauthorized / Forbidden)

• API key was revoked or expired — generate a new one from the provider’s dashboard
• Check that your API key has the necessary permissions for the model you’re using

Context window exceeded

When the conversation exceeds the model’s context limit, the API returns an error. Dino’s garbage collection helps avoid this — it drops processed file contents after each turn while preserving conversation history. Most sessions can run for hundreds of turns before hitting the limit.

If you do:

1. Open a new tab (Cmd+Shift+D) and start a fresh conversation
2. Use Copy as Markdown from the old chat to transfer key context if needed
3. Consider a model with a larger context window for long sessions

Responses are slow

• Larger models (e.g., claude-opus-4-1, gpt-5.5) are slower but more capable
• Use a faster model for quick tasks: gemini-3.5-flash, deepseek-v4-flash
• Coding plans may throttle near daily caps — switch providers
• First request after idle is slower (cold start) — subsequent requests are faster

Workspace Issues

Dino can’t find files

• Open a workspace folder in VS Code, not individual files
• Dino respects .gitignore — ignored files won’t appear in searches
• Very large projects may take a moment on first use

Changes aren’t showing on disk

Dino never writes to disk without your approval. Click Apply to Disk or Apply & Commit in the diff viewer to write changes.

Changes look stale after external edits

If you edited files outside Dino (or another tab applied changes), your staging layer may be out of date. Close and reopen the diff viewer to refresh from disk.

Multiple tabs have conflicting changes

Each Dino tab has its own isolated staging layer. If Tab A applies changes to disk, Tab B will detect the conflict and show a rebase banner. Accept the rebase to merge, or discard Tab B’s changes if they’re no longer needed.

Reset

If Dino is in a broken state, you can reset without losing chat history:

1. Cmd+Shift+P → Dino: Reset Settings and Providers
2. Confirm the reset — this deletes all API keys and preferences
3. Reload the VS Code window
4. Your chat history is preserved

Getting Help

Email support@smartdino.dev — include:

• VS Code version (Help → About)
• Dino extension version (Extensions panel)
• The error message or unexpected behavior
• Steps to reproduce