Troubleshooting
MCP connection issues
Symptom: Your AI tool can’t connect to BaseLayer, or tool calls fail. Fix:- If using remote MCP (Claude.ai, ChatGPT): check your internet connection and try re-authorizing the OAuth connection
- If using IDE MCP (Cursor, Windsurf, etc.): verify your MCP configuration file has the correct URL (
https://api.baselayer.id/mcp) - Try removing and re-adding the BaseLayer MCP integration
- Start a new conversation (MCP tools may not appear in existing ones)
Chrome extension not capturing
Symptom: Conversations aren’t appearing in your memory. Fix:- Click the BaseLayer extension icon and check the status dot
- Green dot: capturing normally
- Gray dot: you’re not on a supported site (Claude, ChatGPT, Gemini, OpenRouter, Open WebUI)
- Red dot: error. Try signing out and back in from the extension popup
- Refresh the AI page and try again
No entities appearing
Symptom: You’ve been using BaseLayer but no entities are showing up. Fix:- Check that your conversations are being captured (look for recent activity in the desktop app)
- Dream engine processing takes time: Pro users ~2 minutes, Free tier up to an hour
- Try asking an AI connected via MCP: “What do you know about me?” to verify the connection
Dream engine not processing
Symptom: Conversations are captured but no entities are being extracted. Fix:- Open the desktop app and check Settings for dream engine status
- On the Free tier, processing runs every 2 hours. Wait for the next cycle
- On Pro, processing should happen within a few minutes. If it’s delayed, try restarting the desktop app
MCP tools not appearing in Claude.ai
Symptom: You connected BaseLayer but Claude doesn’t use the tools. Fix:- Go to Settings → Integrations in Claude.ai and verify BaseLayer is listed
- Try removing and re-adding the integration
- Start a new conversation (tools may not appear in existing ones)
- Try explicitly asking: “Use BaseLayer to search for information about my projects”
Slow search results
Symptom: Queries take a long time to return. Fix:- Use
ask_questionfor direct answers (most efficient) - Use
memory_searchwithdepth: "distilled"for faster results - For
get_entity_relations, use thequeryparameter to filter rather than retrieving all edges