Repository files navigation

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

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
Loki	✅	Query logs for Kubernetes resources or any query
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

HolmesGPT can fetch alerts/tickets to investigate from external systems, then write the analysis back to the source or Slack.

Read theinstallation documentation to learn how to install HolmesGPT.

Read theLLM Providers documentation to learn how to set up your LLM API key.

• In the Robusta SaaS: Go toplatform.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.txtEnter interactive mode to ask follow-up questions:```bashholmes ask"what pods are unhealthy and why?" --interactive# orholmes ask"what pods are unhealthy and why?" -i

Also supported:

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

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

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 commentholmes investigate pagerduty --pagerduty-api-key<PAGERDUTY_API_KEY> --update

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: Seehere
• If using the CLI: Use-t flag withcustom 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 withcustom 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:

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

We donot 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.

Because HolmesGPT relies on LLMs, it relies ona 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).

Distributed under the Apache 2.0 License. SeeLICENSE for more information.

Join our community to discuss the HolmesGPT roadmap and share feedback:

📹First Community Meetup Recording:Watch on YouTube

• Topics: Roadmap discussion, community feedback, and Q&A
• Resources:📝 Meeting Notes |📋 Community Page

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

Please read ourCONTRIBUTING.md for guidelines and instructions.

For help, contact us onSlack or askDeepWiki AI your questions.

Please make sure to follow the CNCF code of conduct -details here.