#help

1 messages · Page 1 of 1 (latest)

maiden surgeBOT Jan 29, 2026, 5:39 PM

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> ✅ Mark Solution

odd lanternBOT Jan 29, 2026, 5:56 PM

I'm seeing this message but I need more context to help troubleshoot! 🦐

When you say "still isn't working," what specifically are you trying to do? And could you share:

Version: moltbot --version (version mismatch is the #1 cause of weird issues!)
What you were setting up (login, model configuration, etc.)
Any error messages you're seeing

This will help me get you unblocked much faster! And remember, the FAQ has solutions for the most common setup hiccups: https://docs.moltbot.ai/start/faq

That "no output agent gemini" error usually points to a configuration issue. Let me help you troubleshoot! 🦐

Quick checks:

Verify your model config - Make sure you have a valid Gemini model specified. Check with:
```
moltbot model list
moltbot model get
```
Check your API key - Ensure your .env or config has the Gemini API key set correctly:
```
GEMINI_API_KEY=your-key-here
```
Try a different model - Sometimes specific model names change. Try:

Ah, I see the full context now! The user (romir.) was having issues starting around 17:39 with a "no output agent gemini" error, and then followed up at 17:41 saying it's still not working.

I've already replied to their 17:41 message with troubleshooting steps. If they respond with their moltbot model get output (like in the screenshot), we can dig deeper!

Common fix for this error: The google/gemini-2.0-flash-exp model name sometimes changes. Ask them to run:

moltbot model set google/gemini-1.5-flash

Or check what Google models are actually available:

moltbot model list | grep google

Let me know if they reply with more details and I can help further! 🦐

Getting "no output" is frustrating! Let's debug this systematically. 🦐

First - critical question: What does moltbot --version show? Version mismatch causes all kinds of weird behavior.

While you check that, try these:

Check if Moltbot is actually running:
```
moltbot gateway status
```
Check the logs for clues:
```
moltbot gateway logs --tail 50
```
Look for any error messages around the time you tried to send a message.

I can see everything above! Here's the summary:

What I know:

You're getting "no output agent gemini" errors
Messages aren't working
You've tried some troubleshooting but it's still broken

What I need to fix this:

moltbot --version - This is crucial! Version mismatch is the #1 cause of these issues
moltbot gateway logs --tail 50 - Shows the actual error behind the scenes

I've asked for these twice now but haven't seen the output yet. Can you run both commands and paste the results? That'll let me see exactly what's wrong instead of guessing! 🦐

While you do that, also check:

Is Moltbot actually running? moltbot gateway status
Did you restart after any config changes? moltbot gateway restart

   moltbot model set google/gemini-2.0-flash-exp

Restart after config changes:
```
moltbot gateway restart
```

Key docs:

Models overview: https://docs.moltbot.ai/concepts/models
Model discussions (community notes): https://docs.moltbot.ai/concepts/models#model-discussions-community-notes

What does moltbot model get show you? That'll help narrow it down!

Try a simple test message:
```
moltbot model set openai/gpt-4o-mini
moltbot gateway restart
```
Does it work with OpenAI? This narrows down if it's a Google-specific issue.
Verify your API key format - Gemini keys should be a specific format. If you copied it from somewhere, make sure no extra spaces snuck in.

The logs will tell us the most! Can you run moltbot gateway logs --tail 100 and paste what you see? That'll show exactly what's going wrong behind the scenes.

Drop the version and logs here and we'll get you sorted!

boreal abyss Jan 30, 2026, 8:17 PM

@odd lantern are there any other tweaks for memory apart from
mem0
supermem
QMD?

odd lanternBOT Jan 30, 2026, 8:33 PM

boreal abyss <@1457407575476801641> are there any other tweaks for memory apart from mem0 sup...

Based on the official docs, here's what's available for memory beyond the three you mentioned:

Built-in Memory Plugins

1. memory-core (default)

• Bundled memory search plugin
• Uses vector search over MEMORY.md and memory/YYYY-MM-DD.md
• Supports hybrid search (BM25 + vector) for both semantic and keyword matching
• Configurable providers: OpenAI, Gemini, or local embeddings
2. memory-lancedb (new!)

• Bundled long-term memory plugin
• Provides auto-recall/capture capabilities
• Set plugins.slots.memory = "memory-lancedb" to switch
Configuration Tweaks

Beyond the plugins, you can tune:

| Option                                  | What it does                                     |
| --------------------------------------- | ------------------------------------------------ |
| `memorySearch.extraPaths`                 | Index additional Markdown directories            |
| `memorySearch.query.hybrid`               | Control vector/text weight blend (default 70/30) |
| `memorySearch.cache.enabled`              | Cache embeddings in SQLite                       |
| `memorySearch.experimental.sessionMemory` | Index session transcripts too                    |
| `memorySearch.store.vector.extensionPath` | Custom sqlite-vec path                           |
| `agents.defaults.compaction.memoryFlush`  | Auto-flush before compaction                     |

Disabling Memory Entirely

Set plugins.slots.memory = "none" to disable all memory plugins.