Solve alerts faster with an AI Agent

How it Works | Installation | LLM Providers | YouTube Demo |

Respond to alerts faster, using AI to automatically:

• Fetch logs, traces, and metrics
• Determine if issues are application or infrastructure related
• Find upstream root-causes

Using HolmesGPT, you can transform your existing alerts from this 👇

To this 👇

How it Works

HolmesGPT connects AI models with live observability data and organizational knowledge. It uses an agentic loop to analyze data from multiple sources and identify possible root causes.

🔗 Data Sources

HolmesGPT integrates with popular observability and cloud platforms. The following data sources ("toolsets") are built-in. Add your own.

Data Source	Status	Notes
ArgoCD	✅	Get status, history and manifests and more of apps, projects and clusters
AWS RDS	✅	Fetch events, instances, slow query logs and more
Confluence	✅	Private runbooks and documentation
Coralogix Logs	✅	Retrieve logs for any resource
Datetime	✅	Date and time-related operations
Docker	✅	Get images, logs, events, history and more
GitHub	🟡 Beta	Remediate alerts by opening pull requests with fixes
DataDog	🟡 Beta	Fetches log data from datadog
Grafana Loki	✅	Query logs for Kubernetes resources or any query
Grafana Tempo	✅	Fetch trace info, debug issues like high latency in application.
Helm	✅	Release status, chart metadata, and values
Internet	✅	Public runbooks, community docs etc
Kafka	✅	Fetch metadata, list consumers and topics or find lagging consumer groups
Kubernetes	✅	Pod logs, K8s events, and resource status (kubectl describe)
NewRelic	🟡 Beta	Investigate alerts, query tracing data
OpenSearch	✅	Query health, shard, and settings related info of one or more clusters
Prometheus	✅	Investigate alerts, query metrics and generate PromQL queries
RabbitMQ	✅	Info about partitions, memory/disk alerts to troubleshoot split-brain scenarios and more
Robusta	✅	Multi-cluster monitoring, historical change data, user-configured runbooks, PromQL graphs and more
Slab	✅	Team knowledge base and runbooks on demand

🔐 Data Privacy

By design, HolmesGPT has read-only access and respects RBAC permissions. It is safe to run in production environments.

We do not train HolmesGPT on your data. Data sent to Robusta SaaS is private to your account.

For extra privacy, bring an API key for your own AI model.

🚀 Bi-Directional Integrations With Your Tools

Robusta can investigate alerts - or just answer questions - from the following sources:

Integration	Status	Notes
Slack	🟡 Beta	Demo. Tag HolmesGPT bot in any Slack message
Prometheus/AlertManager	✅	Robusta SaaS or HolmesGPT CLI
PagerDuty	✅	HolmesGPT CLI only
OpsGenie	✅	HolmesGPT CLI only
Jira	✅	HolmesGPT CLI only

See it in Action

Installation

You can install HolmesGPT in one of the follow three methods:

1. Standalone: Run HolmesGPT from your terminal as a CLI tool. Typically installed with Homebrew or Pip/Pipx. Ideal for local use, embedding into shell scripts, or CI/CD pipelines. (E.g. to analyze why a pipeline deploying to Kubernetes failed.)
2. Web UIs and TUIs: HolmesGPT is embedded in several third-party tools, like Robusta SaaS and K9s (as a plugin).
3. API: Embed HolmesGPT in your own app to quickly add root-cause-analysis functionality and data correlations across multiple sources like logs, metrics, and events. HolmesGPT exposes an HTTP API and Python SDK, as well as Helm chart to deploy the HTTP server on Kubernetes.

Standalone Brew Pipx Docker Docker Build Poetry
Web UIs and TUIs Robusta SaaS K9s Plugin	API Helm Chart Python API

Supported LLM Providers

Select your LLM provider to see how to set up your API Key.

OpenAI	Anthropic	AWS Bedrock	Azure OpenAI	Google Vertex AI
Gemini	Ollama

You can also use any OpenAI-compatible models, read here for instructions.

Using HolmesGPT

• In the Robusta SaaS: Go to platform.robusta.dev and use Holmes from your browser
• With HolmesGPT CLI: setup an LLM API key and ask Holmes a question 👇

holmes ask "what pods are unhealthy and why?"

You can also provide files as context:

holmes ask "summarize the key points in this document" -f ./mydocument.txt

You can also load the prompt from a file using the --prompt-file option:

holmes ask --prompt-file ~/long-prompt.txt

Enter interactive mode to ask follow-up questions:
```bash
holmes ask "what pods are unhealthy and why?" --interactive
# or
holmes ask "what pods are unhealthy and why?" -i

Also supported:

HolmesGPT CLI: investigate Prometheus alerts

Pull alerts from AlertManager and investigate them with HolmesGPT:

holmes investigate alertmanager --alertmanager-url http://localhost:9093
# if on Mac OS and using the Holmes Docker image👇
#  holmes investigate alertmanager --alertmanager-url http://docker.for.mac.localhost:9093

To investigate alerts in your browser, sign up for a free trial of Robusta SaaS.

Optional: port-forward to AlertManager before running the command mentioned above (if running Prometheus inside Kubernetes)

kubectl port-forward alertmanager-robusta-kube-prometheus-st-alertmanager-0 9093:9093 &

HolmesGPT CLI: investigate PagerDuty and OpsGenie alerts

holmes investigate opsgenie --opsgenie-api-key <OPSGENIE_API_KEY>
holmes investigate pagerduty --pagerduty-api-key <PAGERDUTY_API_KEY>
# to write the analysis back to the incident as a comment
holmes investigate pagerduty --pagerduty-api-key <PAGERDUTY_API_KEY> --update

For more details, run holmes investigate <source> --help

Customizing HolmesGPT

HolmesGPT can investigate many issues out of the box, with no customization or training. Optionally, you can extend Holmes to improve results:

Custom Data Sources: Add data sources (toolsets) to improve investigations

• If using Robusta SaaS: See Robusta's docs
• If using the CLI: Use -t flag with custom toolset files or add to ~/.holmes/config.yaml

Custom Runbooks: Give HolmesGPT instructions for known alerts:

• If using Robusta SaaS: Use the Robusta UI to add runbooks
• If using the CLI: Use -r flag with custom runbook files or add to ~/.holmes/config.yaml

You can save common settings and API Keys in a config file to avoid passing them from the CLI each time:

Reading settings from a config file

You can save common settings and API keys in config file for re-use. Place the config file in ~/.holmes/config.yaml` or pass it using the --config

You can view an example config file with all available settings here.

Evals

Because HolmesGPT relies on LLMs, it relies on a suite of pytest based evaluations to ensure the prompt and HolmesGPT's default set of tools work as expected with LLMs.

• Introduction to HolmesGPT's evals.
• Write your own evals.
• Use Braintrust to view analyze results (optional).

License

Distributed under the MIT License. See LICENSE.txt for more information.

Support

If you have any questions, feel free to message us on robustacommunity.slack.com

How to Contribute

Install HolmesGPT from source with Poetry. See Installation for details.

For help, contact us on Slack or ask DeepWiki AI your questions.