Describe your data backend
in plain language.
Let an AI agent build it for you. The dcupl CLI is designed to be driven by agents — paired with the dcupl skill in Claude Code, an agent can scaffold projects, ingest and profile data, author models and workflows, and deploy to the cloud, all from your terminal. You stay in the loop, the agent does the keystrokes, and every step is inspectable.
you: "what's in products.csv, and what's the price distribution?": dcupl app create --auto-update --json — output: { "appId": "swift-otter-3a1f", "port": 52341 }
dcupl app data upsert --file products.csv --model Product \ --key-property id --auto-generate-sample-size 100 --json — output: { "upserted": 1284, "model": "Product", "built": true }
dcupl app fn aggregate --model Product --attribute price --types avg,min,max --json — output: { "attribute": "price", "types": ["avg", "min", "max"], "avg": 84.27, "min": 4.99, "max": 489 }
The three pillars
The CLI was built to be agent-ready. Three pillars make the difference — Why it works breaks down the CLI design choices behind them.
Machine-readable output
With --json, the app, validate, workflow, files and schemas commands answer in structured JSON, errors included. Instead of scraping human-formatted text, the agent gets output it can parse reliably — so it knows exactly what happened and what to do next.
--jsonAn explicit lifecycle
Data mutations don't apply silently. They stage changes that you apply with dcupl app build (or immediately with --auto-update). Each step is predictable and inspectable, so the agent can reason about state and recover cleanly.
dcupl app buildThe dcupl skill
A Claude Code plugin that encodes dcupl domain knowledge: command patterns, the right entry points for each task, and the non-obvious gotchas. It turns the CLI from a tool the agent might misuse into one it drives correctly.
How it fits together
You talk to Claude Code in plain language. The skill tells the agent how to use the CLI. The CLI stages, builds, and syncs your project against the dcupl cloud — and the structured output flows back so the agent can verify each step.