Opsfile

Installation

curl -fsSL https://seanseannery.github.io/opsfile/install.sh | bash

brew tap seanseannery/opsfile https://github.com/seanseannery/opsfile
brew install seanseannery/opsfile/opsfile

npm install -g github:seanseannery/opsfile

Requires the ops CLI to be installed first — use curl, Homebrew, or npm above.

/plugin marketplace add seanseannery/opsfile
/plugin install ops@opsfile

Getting Started

Step 1: Create an Opsfile

Create an Opsfile in your repo root. It uses Makefile-style syntax:

# Variables — prefix with environment name to scope them
prod_AWS_ACCOUNT=1234567
preprod_AWS_ACCOUNT=8765431

# Commands — define per-environment shell lines
# Use "default" as a fallback when env-specific block is absent
tail-logs:
    default:
        aws logs tail /aws/ecs/my-service --follow --since 30m
    local:
        docker logs myapp --follow --tail 100

list-instance-ips:
    prod:
        aws ec2 describe-instances --filters "Name=tag:Env,Values=prod" --query "Reservations[*].Instances[*].PrivateIpAddress" --output text
    preprod:
        aws ecs list-container-instances --cluster my-cluster-preprod --output text

# Shell env vars are injected automatically — ops falls back to env if not in Opsfile
show-profile:
    default:
        echo "Using AWS profile: $(AWS_PROFILE)"

or use the Claude plugin

Run /ops:init-opsfile to have Claude analyze your repo's tech stack and generate a tailored Opsfile automatically.

$ claude "/ops:init-opsfile"

Step 2: Run ops commands

Run commands with the following syntax:

$ ops [flags] <environment> <command> [args]

Flag	Short	Description
`--directory <path>`	`-D <path>`	Use the Opsfile in the given directory
`--env-file <path>`	`-e <path>`	Load secrets from a `.env` file (defaults to `.ops_secrets.env`)
`--dry-run`	`-d`	Print resolved commands without executing
`--list`	`-l`	List available commands and environments
`--silent`	`-s`	Execute commands without printing output
`--version`	`-v`	Print the ops version and exit
`--help`	`-h`	Show usage information and exit

Examples:

$ ops preprod instance-count

$ ops prod open-dashboard

$ ops local logs

$ ops prod k8s -namespace myspace

or use the Claude plugin

Pass a command directly or describe what you need in plain English — /ops:run discovers your available commands and resolves the best match.

$ claude "/ops:run prod tail-logs"

$ claude "/ops:run --list"

$ claude "/ops:run are there any errors in staging?"

Why use it?

⚡

Less Stress

Quickly find the right command during a live outage

🤖

Less AI tokens

CLI > MCP. Agents run ops scripts instead of googling commands or loading large MCP context

📖

Knowledge Sharing

Share runbooks with your team as code

📦

Encapsulation

Keep your Makefile focused on CI/CD

🤝

Team Standardisation

No more tribal knowledge or bespoke bash aliases. Opsfile lives in the repo.

🔒

Secrets-Safe by Design

Inject secrets from .env files with -e — never commit them to your Opsfile.