"Create a workflow that monitors GitHub PRs and posts to Slack when they're approved"
I've created a workflow with 3 nodes: GitHub webhook trigger, approval filter, and Slack notification. You can drag to reposition or click nodes to configure.
GitHub webhook trigger
What starts this workflow.
- Eventpull request review
- Repositoryyour main repo
- Fires whena review is submitted
- AuthenticationGitHub app token
In the full version you pick the repository and event from a list, and test the trigger before saving.
Filter: approved
What gets through to the next step.
- Conditionreview state is approved
- Passed last week38 of 61
- Blockedcomments and change requests
- Runsin under 50ms
Conditions are plain rules you can read. No formula language to learn.
Slack message
What the workflow does at the end.
- Channel#engineering
- MessagePR title, author, link
- Sent last week38
- Failures0
The same step can post to Teams or send an email instead. You choose when you build it.
Two ways to work on the same thing
Visual builder and code, kept in step.
- Edit visuallyanyone on the team
- Edit the codedevelopers
- Syncboth directions
- Exportplain files you own
A change in one view shows up in the other. Nothing is trapped in a builder.
Pick the model per task
Five providers, chosen by the job.
- Cheap and fastsmall models
- Hard reasoninglarge models
- Private workself-hosted models
- Switching costnone
Routing work to the right model is where the savings come from, not from a cheaper subscription.
Built-in integrations
The systems a workflow can reach.
- IncludedGitHub, Slack, PostgreSQL, Stripe
- Also includedSendGrid, Zapier, Twilio, AWS S3, Docker, Jira
- Community extensionssupported
- Custom integrationssupported
Each integration is a connection you approve. Nothing reaches out until you connect it.
Built for a whole team
Who does what.
- Productdescribes the workflow
- Designshapes the steps
- QAtests it
- Developersedit the code behind it
Non-developers use the visual mode. Developers are never boxed out of the code.
Run it where you need it
Deployment options.
- Dockeron your own server
- On premisesupported
- Air gappedsupported
- Your data leavesonly where you connect it
Self hosting is there for regulated work. Most small teams will not need it.
Nothing is locked in
What you can take with you.
- Workflowsexported as files
- Codeyours, plain
- Dataexported any time
- Proprietary formatsnone
If you stop using it, the work you built still runs.
What Makes Horns Code Different
The only platform combining visual workflows with full code access
Dual-Mode Interface
Switch between visual workflow builder and code mode instantly. Bi-directional sync keeps both in perfect sync - an industry first.
Multi-LLM Cost Optimizer
5 providers: Gemini, GPT, Claude, Llama, DeepSeek. Send the simple work to a cheap model and the hard work to a strong one, instead of paying one rate for everything.
MCP Marketplace
Built in: GitHub, Slack, PostgreSQL, Stripe, SendGrid, Zapier, Twilio, AWS S3, Docker, Jira. Plus unlimited community extensions.
Multi-Persona Platform
Product Managers, Designers, QA, Developers, Analysts - everyone on your team can contribute. Visual mode for non-coders, code mode for devs.
Self-Hosted & Secure
Docker deployment, on-premise, air-gapped options. Full data sovereignty for HIPAA, SOX, and government compliance requirements.
Zero Vendor Lock-in
Export everything - full code ownership with no proprietary formats. Your workflows, your code, your data. Leave anytime with complete portability.
Unlike single-model tools: Switch AI providers based on task complexity and budget. Unlike no-code platforms: Full code transparency - see and edit generated code anytime. Unlike cloud-only solutions: Self-hosted option keeps sensitive data in your infrastructure.
Choose Your AI Model
Switch between leading LLMs based on your needs
Gemini Flash
Fast multimodal model • 1M context
GPT
Complex reasoning • 128K context
Claude Sonnet
Nuanced coding • 200K context
Llama
Run locally via Ollama • Complete privacy • Free
DeepSeek
Cost-effective • 64K context
Flexible Pricing
Self-hosted is free forever. Cloud plans scale with your team.
Self-Hosted
- All 5 AI models
- Visual + Code modes
- MCP marketplace
- Community support
Pro
- Unlimited requests
- All AI models
- Priority support
- Custom prompts
Team
- Everything in Pro
- Shared workspace
- WebSocket collab
- Admin controls
Enterprise
- SSO / SAML
- HIPAA / SOX ready
- Air-gapped option
- SLA guarantee
Why several models: simple work goes to a cheap, fast model and hard work goes to a strong one, so you are not paying one rate for everything.