> ## Documentation Index
> Fetch the complete documentation index at: https://docs.getconcierge.app/llms.txt
> Use this file to discover all available pages before exploring further.

# Logs

> View logs from your deployed servers

Real-time and historical logs for servers deployed with `concierge deploy`. Use it to debug startup issues, trace errors, and monitor server health.

## What You See

### Startup Logs

When your server boots, Concierge logs:

* Python version and dependencies loaded
* Registered tools, resources, and prompts
* Stage and transition configuration
* Transport binding (port, protocol)
* State backend connection status

```
[2026-03-08 10:23:01] INFO  Starting Concierge server "shopping"
[2026-03-08 10:23:01] INFO  Registered 6 tools: search_products, add_to_cart, ...
[2026-03-08 10:23:01] INFO  Stages: browse → cart → checkout
[2026-03-08 10:23:01] INFO  State backend: postgres (connected)
[2026-03-08 10:23:01] INFO  Listening on https://your-server.concierge.app
```

### Runtime Errors

Errors originating from your tool code are captured with full tracebacks:

```
[2026-03-08 10:25:33] ERROR Tool "checkout" raised exception:
  File "main.py", line 42, in checkout
    total = calculate_total(cart)
  TypeError: calculate_total() got an unexpected keyword argument 'currency'
```

### Request Logs

Every MCP request/response is logged:

```
[2026-03-08 10:24:15] INFO  → tools/call: search_products(query="laptop")
[2026-03-08 10:24:15] INFO  ← 200 OK (23ms): {products: [...]}
[2026-03-08 10:24:18] INFO  → tools/call: proceed_to_next_stage(target="cart")
[2026-03-08 10:24:18] INFO  ← 200 OK (2ms): stage changed to "cart"
```

## Accessing Logs

### CLI

```bash theme={null}
# Stream live logs from current project directory
concierge logs

# Stream logs for a specific project
concierge logs <project_id>
```

### Dashboard

Logs are also available in the [Concierge Platform](https://getconcierge.app) dashboard, with:

* Full-text search
* Log level filtering (DEBUG, INFO, WARN, ERROR)
* Time range selection
* Session ID filtering

<Tip>
  Use session ID filtering to trace a single user's entire workflow:from the first tool call to checkout.
</Tip>