Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/wack/canary

MultiTool Canary is your AI-powered, agentic deployment solution for seamless, risk-managed rollouts
https://github.com/wack/canary

agentic-ai canary-release deployment-automation

Last synced: 3 days ago
JSON representation

MultiTool Canary is your AI-powered, agentic deployment solution for seamless, risk-managed rollouts

Awesome Lists containing this project

README

        

multitool-canary-github

MultiTool Canary


Your AI-powered, agentic deployment solution for seamless, risk-managed updates


🏡 MultiTool website | 🗪 Contact us

## What is MultiTool Canary?

MultiTool Canary is an agentic deployment tool that quickly detects and rolls back defective deployments. Use the CLI to create and monitor canary deployments, control the amount of traffic they receive, and automatically yank deployments that produce an anomolous number of bad responses. MultiTool Canary improves release confidence by giving you a robust measure of stability, and acting quickly and autonomously when bugs sneak into production.

## Getting Started

This section explains how to deploy an AWS Lambda function to AWS API Gateway using MultiTool Canary. Just point MultiTool Canary to a pre-built Lambda zip, tell it which API Gateway, Stage, and Lambda to deploy to, and it will take care of the rest.

1. First, make sure your AWS credentials are configured:
```bash
$ aws sts get-caller-identity
```

If your credentials aren't set, either login with `aws sso login --profile my-profile` or set your Access Key Id and Secret Access Key:
```bash
$ export AWS_ACCESS_KEY_ID="${MY_ACCESS_KEY}"
$ export AWS_SECRET_ACCESS_KEY="${MY_SECRET_ACCESS_KEY}"
$ export AWS_REGION="us-east-2"
```

2. Next, tell MultiTool Canary which API Gateway, Stage, and Lambda it should deploy it.
```bash
$ export MULTI_GATEWAY_NAME="${MY_API_GATEWAY_NAME}"
$ export MULTI_GATEWAY_STAGE="${MY_STAGE}"
$ export MUTLI_LAMBDA_NAME="${MY_LAMBDA}"
```

3. Build your Lambda function. This should be a zipfile with your Lambda's source code. For instance, if you're using `cargo-lambda`:
```bash
$ cargo lambda build -o zip --release
$ mv ./target/lambda/my-app/bootstrap.zip ./my-release.zip
```

4. Finally, run MultiTool Canary, pointing to a pre-built zip:
```bash
$ canary deploy ./my-release.zip
```

MultiTool Canary will upload the function to AWS, cut a canary release, and monitor it, progresssingly ramping up traffic as it gains confidence in the deployment. MultiTool Canary scans your CloudWatch metrics to see how your canary application is performing relative to the baseline, and models release confidence based on how they differ.

## Table of Contents

- [What is MultiTool Canary?](#what-is-multitool-canary-)
- [Getting Started](#getting-started)
- [Table of Contents](#table-of-contents)
- [Installation](#installation)
* [Building from source](#building-from-source)
- [Architecture support](#architecture-support)
- [Mission](#mission)
- [Contributing](#contributing)

## Installation

Install MultiTool Canary with Homebrew, NPM, Powershell, or via `curl`.

**Installing with Homebrew:**

```sh
brew install wack/tap/canary
```

**Installing with NPM:**

```sh
npm install @multitool/[email protected]
```

**Installing with curl:**

```sh
curl --proto '=https' --tlsv1.2 -LsSf https://github.com/wack/canary/releases/download/v0.1.0-alpha.1/canary-installer.sh | sh
```

**Installing with Powershell:**

```sh
powershell -ExecutionPolicy ByPass -c "irm https://github.com/wack/canary/releases/download/v0.1.0-alpha.1/canary-installer.ps1 | iex"
```

See [Releases](https://github.com/wack/canary/releases) for pre-built binaries, checksums, and guides to install on additional platforms.

### Building from source

To build MultiTool Canary from source, refer to instructions in [CONTRIBUTING.md](https://github.com/wack/canary/blob/trunk/README.md).

## Features

* Automatic application deployment
* Automated rollback
* Cautious Mode (coming soon) and Optimistic Mode: Choose between preventing false positives and false negatives
* Measure release confidence
* Backward and forward traffic ramping

## Platform Support

Currently, MultiTool Canary only supports deploying canaries for AWS Lambda functions running in AWS API Gateway. However, we are currently working on adding support for additional platforms. Our platform roadmap is presented in the table below.

| Platform | Support |
| :----------------------------------------------------------: | :----------------------------: |
| **AWS Lambda + API Gateway** | :sparkles: Available! |
| [Vercel](vercel.com) | :eyes: Up next |
| [CloudFlare](https://developers.cloudflare.com/workers/configuration/versions-and-deployments/gradual-deployments/#_top) | :eyes: Up next |
| [AWS Function Aliases](https://docs.aws.amazon.com/AWSCloudFormation/latest/UserGuide/aws-resource-lambda-alias.html) | :watch: Soon |
| [Kubernetes](https://kubernetes.io/) | :hourglass_flowing_sand: Later |
| [Google Cloud Run Functions](https://cloud.google.com/functions) | :hourglass_flowing_sand: Later |

## Architecture support

MultiTool Canary has official builts for 64-bit Windows, macOS (Apple Silicon and x86), and Linux. No official support is provided for 32-bit architectures, but the CLI is expected to work on 32-bit platforms.

| | Apple Silicon macOS | Intel macOS | x64 Windows | x64 Linux |
| :--------: | :-----------------: | :---------: | :---------: | :-------: |
| Homebrew | ✓ | ✓ | ✓ | ✓ |
| NPM | ✓ | ✓ | ✓ | ✓ |
| Curl | ✓ | ✓ | ✓ | ✓ |
| Powershell | ✓ | ✓ | ✓ | ✓ |

## Mission

We're building tools to take agentic deployments mainstream. Right now, operators must synchronously watch deployments go live, or rely on passive metric alerting to identify service disruptions. We want to empower operators to *proactively* identify and rollback disruptive deployments *before* they reach 100% of users with the help of an AI agent. MultiTool Canary is our foundational strategy to introduce agentic deployments to everyone. To learn more about us, visit [our company website](https://www.multitool.run/company).

## Contributing

:rocket: We accept pull requests! :guitar:

See [CONTRIBUTING.md](https://github.com/wack/canary/blob/trunk/CONTRIBUTING.md) for more information.