Skip to content
SWEny

Quick Start

SWEny is one tool with multiple surfaces. Install it once, then pick the way that fits your workflow.

Install

Section titled “Install”
Terminal window
npm install -g @sweny-ai/core

You can also run it directly with npx @sweny-ai/core.

Add your API key

Section titled “Add your API key”

SWEny uses Claude as its AI engine. You’ll need an Anthropic API key, OAuth token, or authenticated Claude Code instance.

Terminal window
# .env (gitignored)
ANTHROPIC_API_KEY=sk-ant-...

Or use a Claude subscription token (CLAUDE_CODE_OAUTH_TOKEN) for flat-rate billing. The CLI auto-loads .env at startup.

Choose your surface

Section titled “Choose your surface”

Claude Code Plugin — use SWEny inside Claude Code

Section titled “Claude Code Plugin — use SWEny inside Claude Code”

If you use Claude Code, install the plugin and get 11 slash commands, MCP tools, and a startup hook:

/plugin marketplace add swenyai/sweny
/plugin install sweny@sweny-official

Then use /sweny:triage to investigate production alerts, /sweny:implement ENG-123 to fix an issue, /sweny:e2e-run to run browser tests, or /sweny:workflow-create to build a custom workflow — all without leaving your conversation.

Full plugin guide — all skills, MCP tools, hooks, and agent details.

CLI — build and run workflows from your terminal

Section titled “CLI — build and run workflows from your terminal”

The fastest way to get things done. Describe a task, get a workflow, run it.

Terminal window
# Create a workflow from a description
sweny workflow create "scan the codebase for security anti-patterns \
  and create tickets for critical findings"


# Refine it
sweny workflow edit .sweny/workflows/security_scan.yml \
  "add a quality gate that rejects vague findings"


# Run it
sweny workflow run .sweny/workflows/security_scan.yml

You can also run the built-in workflows directly:

Terminal window
sweny triage --dry-run          # investigate production errors
sweny implement ENG-123         # fix a tracked issue and open a PR

Or generate AI-driven browser tests for any web app:

Terminal window
sweny e2e init                  # wizard generates test workflows
sweny e2e run                   # AI agent drives a real browser

Full CLI guide — commands, configuration, and real-world examples. E2E testing guide — browser test generation and execution.

GitHub Action — deploy workflows to CI

Section titled “GitHub Action — deploy workflows to CI”

Put workflows on a schedule. SWEny monitors your observability platform, triages errors, and opens fix PRs — automatically.

.github/workflows/sweny-triage.yml
- uses: swenyai/triage@v1
  with:
    claude-oauth-token: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
    observability-provider: sentry
    sentry-auth-token: ${{ secrets.SENTRY_AUTH_TOKEN }}
    sentry-org: my-org
    sentry-project: my-project

Three secrets. Push the file, trigger it from the Actions tab, and check the summary. Use swenyai/sweny@v5 to run custom workflow YAMLs, or swenyai/e2e@v1 for browser tests.

Full Action guide — setup, inputs, scheduling, and service maps.

Studio — visualize and monitor workflows

Section titled “Studio — visualize and monitor workflows”

A visual DAG editor and live execution monitor built on React Flow. Design workflows by dragging nodes, or watch running workflows execute node-by-node in real time.

Terminal window
# Stream a CLI run to Studio
sweny triage --stream

Full Studio guide — editor, embedding, and live mode.

Marketplace — start from a community workflow

Section titled “Marketplace — start from a community workflow”

Don’t want to write a workflow from scratch? Browse marketplace.sweny.ai for ready-to-run SWEny workflows. Pick one, copy the YAML into .sweny/workflows/, customize the steps, and run.

What’s next?

Section titled “What’s next?”