> ## Documentation Index
> Fetch the complete documentation index at: https://docs.ewake.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Interacting with the Agent
> How your team interacts with ewake in day-to-day use, on-demand questions, automated responses, and slash commands.
Slack is where most teams interact with ewake every day. Whether it's asking a question during an incident, receiving an automated alert investigation, or reviewing a scheduled health report, everything arrives in Slack threads, in channels your team already uses.
For configuration, use the [Dashboard](/working-with-ewake/dashboard).
***
## How to interact with ewake
### On-demand
Mention `@ewake` followed by your question in any channel where the bot is present:
```
@ewake any anomalies in the past 6 hours?
```
```
@ewake what changed in the last deploy on checkout-service?
```
Replace `checkout-service` with the name of your service as it appears in Datadog.
Ewake will reply in a thread with a structured investigation response.
### Automated responses
Ewake can respond automatically, no mention needed, when a use case is configured.
**On-Call Agent**
Ewake monitors a configured Slack channel and automatically replies in the alert thread whenever a configured alert bot posts. Delivers probable cause, evidence, and suggested next steps before your engineer opens a terminal.
**Incident Response**
During a P0 or war room, mention `@ewake` directly in the incident channel to run a real-time investigation alongside your team.
**Scheduled Tasks**
Ewake posts structured reports to your configured channel at the scheduled time, health briefs, anomaly summaries, and reliability digests.
### Slash commands
Trigger ewake actions directly from the Slack message bar:
| Command | What it does |
| -------------------- | ------------------------------------------------ |
| `/ewake investigate` | Start a new investigation in the current channel |
| `/ewake snooze` | Snooze alert responses for a defined window |
Confirm the exact command names with your ewake workspace, available commands may vary by configuration.
***
## Behavior
### Threaded replies
Ewake always responds in a thread, not in the main channel. When you continue a conversation in the same thread, ewake follows the thread context, you don't need to repeat the full question.
### Direct messages
You can send questions to ewake directly via DM. Responses follow the same format as channel interactions. DM investigations don't have access to channel-specific configuration.
### Follow-up questions
Ewake remembers the context of the current thread. Once it has replied, your engineer can ask follow-up questions directly in the same thread without repeating the full context.
```
@ewake tell me more about evidence #2
```
```
@ewake what would you check next?
```
Ewake uses the full thread history to answer, so investigations can deepen progressively without starting over.
### Feedback
Every ewake response includes a thumbs up / thumbs down reaction. Use it to signal whether the investigation was useful.
Feedback is used to improve ewake's hypotheses over time. A thumbs down on a wrong root cause helps ewake weight similar signals differently in future investigations on your environment.
### Channel-specific configuration
Each channel where ewake is active can have its own instruction prompt, defining the tone, depth, and focus of ewake's responses. Configure this in the ewake dashboard under **Alert Response** → select your trigger → edit the instruction prompt.
***