Skip to main content
Connic
Getting Started

Quickstart

A first Connic project, from setup and agent deployment to the first run through a connector.

Last updated
On this page
Set up with a coding agent

Choose a coding client, install the Connic plugin when a native package is available, then describe the intended build to the agent. The same guide covers separate skill and MCP setup for other clients. Follow Coding agent setup.

  1. Create a Project in Connic

    Create a new project in the Connic dashboard. The wizard guides the repository connection.

  2. Set Up the Local Project

    Install the SDK, then either scaffold from a template or create a minimal project.

    terminal
    pip install connic-composer-sdk

    Option A: Start from a template (recommended). Browse agent templates and run:

    terminal
    connic init my-project --templates=invoice,customer-support --skill
cd my-project

    Option B: Start from scratch. Creates a minimal structure:

    terminal
    connic init my-agents --skill
cd my-agents

    The optional --skill flag adds the Connic skill to the project. In an interactive terminal, the CLI also detects Codex and Claude Code and offers to install the full plugin, which bundles the skill and MCP. Decline the prompt to keep the skill-only setup. See Coding agent setup.

  3. Define the Agent

    Edit the agent YAML file to configure the agent.

    agents/assistant.yaml
    version: "1.0"

name: assistant
model: connic/gpt-5.6-terra
description: "A helpful assistant"
system_prompt: |
  You are a helpful assistant. Answer questions
  clearly and concisely.

temperature: 0.7
max_concurrent_runs: 1

    Each YAML file in the agents/ folder defines one agent. A project can contain multiple agents. For simple projects, keep them at the top level; for larger projects, organize agents into subfolders under agents/.

  4. Deploy

    Choose Git (auto-deploy on push) or CLI (works with any provider).

    Git integration

    1. Connect the repository in Project Settings → Git & Environments

    2. Map each environment to a branch on the same page

    3. Push to trigger deployment

    terminal
    git add .
git commit -m "Initial agent setup"
git push origin main
    CLI deploy

    1. Create API key in Project Settings → API Keys & MCP Auth

    2. Run connic login in the project folder

    3. connic dev to choose a quick test or a reusable named environment. Use connic deploy to select a deployment target by name and confirm.

    terminal
    connic login
connic deploy

  5. Create a Connector

    Create a connector from the marketplace to connect the deployed agent to real traffic. For example, an HTTP webhook can add an AI agent to an existing product through API calls.

    1. Open the agent detail page in the Agents section

    2. Click the + button on the connector flow diagram, then choose Create New Connector

    3. Pick a connector type from the marketplace and configure it (name, mode, etc.)

    4. The connector is automatically linked to the agent once created

  6. Trigger and Inspect the Agent

    Open Playground from the project navigation and select an active agent. For a text, tool, or sequential agent, enter Text or JSON input and click Run. You can attach up to 10 files with a combined size of 10 MiB or schedule the run up to 30 days ahead. If the agent has deployed test variants, choose the base agent or a variant under Run version. A voice agent starts a live microphone session instead; select its base version or a deployed voice variant and click Run. Playground uses the selected environment's tools and connections and shows the output, traces, logs, and recent manual runs.

    You can also trigger the agent through its connector. For an HTTP webhook connector, use a curl command:

    terminal
    curl -X POST <webhook-url> \
  -H "Content-Type: application/json" \
  -H "X-Connic-Secret: <your-secret-key>" \
  -d '{"message": "Hello, agent!"}'

    Check the Runs section in the project to see the agent response and trace details.

Setup complete

The agent is now deployed and ready to use.

Keep building

Project Structure

Understand how Connic Composer SDK works

Agent Configuration

Configure LLM, Sequential, and Tool agents

Tools

Create Python functions that agents can call

Connectors

Integrate agents with external systems