Skip to main content

​
OpenSRE CLI Setup

Set up or maintain opensre with minimal friction. If you are running this prompt, follow the steps below and set up OpenSRE for the user so they can investigate incidents from the CLI or the interactive shell. Do not clone the OpenSRE repo unless the user asked to contribute. The published CLI is enough.

​
Step 1: Install or upgrade the CLI

Check whether the CLI exists:
On Windows PowerShell:
If opensre is missing, install with one of these methods. Prefer the one-line installer unless the user already uses Homebrew.
  1. macOS / Linux (one-line installer):
OPENSRE_AUTO_LAUNCH=0 skips auto-starting the onboard wizard so you can run the remaining steps yourself.
  1. macOS (Homebrew):
  1. Windows (PowerShell):
No sudo on macOS/Linux in the usual case. If the installer cannot use a writable directory already on PATH, it puts the binary in ~/.local/bin and prints the command to add that directory. Apply that PATH update (or open a new terminal), then re-check command -v opensre. When opensre is already present, upgrade it:
Homebrew installs can also use:
Confirm it runs:

​
Step 2: Onboard

Onboard is interactive and needs a TTY. Do not try to fake the wizard. Run it and prompt the user when it asks for input:
The wizard asks for:
  1. An LLM — OpenAI, Anthropic, a local model (Ollama), or another provider they already use. They need an API key, or they can leave the key blank and add it later with opensre auth login <provider>.
  2. Tools — whatever investigations should query (Datadog, Grafana, Slack, and so on). Skip any they do not have.
  3. Optional extras — masking and messaging.
Zero-config local LLM (Ollama, no API key):
To add or change only the LLM later:
To connect one tool later:
Replace <service> with a slug such as datadog, grafana, or slack. If onboard cannot reach the LLM provider (firewall, proxy, offline), tell the user they can Save anyway without validating and continue. If it cannot persist the key, they can Continue without saving (this session only) and re-enter it next time.

​
Step 3: Verify

One tool only:
If verify fails, check the printed error. Common causes: missing or expired credential, wrong URL, or the tool was skipped during onboard. Re-run opensre integrations setup <service> for that tool.

​
Step 4: Suggest a first run

Prefer the interactive shell unless the user already has an alert file:
That starts a TTY REPL. Ask the user to describe an incident in plain language, or use /help. If they have an OpenSRE git checkout, they can run the sample alert from the repo root:
That fixture path does not exist for a binary-only install. Do not invent a sample file. If they already have an alert JSON:

​
Gotchas

  • opensre: command not found — new terminal, or add the bin directory the installer printed (often ~/.local/bin on macOS/Linux).
  • Onboard blocks or looks hung — it is waiting on the user. Show them the prompt; do not kill it.
  • Installer started onboard on its own — that is expected without OPENSRE_AUTO_LAUNCH=0. Let the user finish it, then continue from Step 3.
  • Investigations need a configured LLM — opensre investigate fails at startup without LLM_PROVIDER and a matching key. Finish onboard or opensre auth login <provider> first.
  • Only connected tools are queried — a Datadog alert cannot be investigated if Datadog was never set up. Run opensre integrations verify before a production run.
Human docs: https://opensre.com/docs/install and https://opensre.com/docs/quickstart