Troubleshooting

The troubleshooting engine covers 11 network domains across multiple vendors. Describe the problem in plain English, and the assistant walks you through diagnosis using vendor-specific documentation and proven methodology.

Advisor Mode

By default, Damira AI runs in advisor mode. This means:

• The assistant recommends commands for you to run on your devices
• You paste the output back into the chat
• The assistant analyzes the output and continues diagnosis
• No SSH connections. No credentials stored. Your data stays on your network.

This is the most secure operating mode. The assistant never touches your devices directly.

Supported Domains

How It Works

1. You describe the problem — "OSPF adjacency won't come up between core-rtr-01 and dist-rtr-02"
2. Domain identification — The assistant identifies this as an OSPF troubleshooting request
3. Monitoring check — If you have Prometheus, Grafana, or Loki connected, the assistant queries your monitoring systems first to pull metrics, alerts, and logs related to the issue
4. Documentation lookup — Relevant vendor documentation is pulled automatically
5. Structured diagnosis — The assistant leads with what your monitoring shows, then recommends specific commands for anything monitoring doesn't cover
6. Iterative analysis — You paste command output, the assistant narrows down the root cause

The assistant follows a structured diagnostic methodology: Gather → Isolate → Diagnose → Recommend → Prevent. It always checks your monitoring tools before asking you to run commands manually.

Monitoring-Assisted Troubleshooting

When Prometheus, Grafana, or Loki are connected (see Monitoring Integration), the assistant automatically:

• Queries Prometheus for device metrics, BGP peer state, interface counters, and active alerts
• Checks Grafana for firing alert rules related to the affected device
• Searches Loki for recent log entries matching the device or protocol

This means you often get a diagnosis before running a single command. The assistant tells you what it found from monitoring and only asks for CLI output when it needs data your monitoring doesn't capture.

Tips for Better Results

• Be specific about the problem — "BGP neighbor flapping every 30 seconds" is better than "BGP is broken"
• Include device context — Mention the vendor and platform if possible (e.g., "Cisco IOS-XE 17.x", "Palo Alto PA-5200")
• Paste full command output — The assistant parses structured output from show commands, so include everything
• Use multi-turn conversation — The assistant remembers context within a session. Follow up with additional details as you gather them.

Config Security Audit

Paste a device configuration into the chat with a request like:

Audit this config for security issues:

hostname core-rtr-01
...

The assistant checks for common security gaps: default credentials, unencrypted protocols, missing ACLs, open management interfaces, and more.

Config Generation

Describe what you need and the assistant generates configuration:

• Subnet calculations and IP addressing
• Interface configurations
• Routing protocol setup
• Access control lists