Quickstart

Quickstart

Go from zero to a completed agent task in under five minutes. You will create an account, add your AI provider key, load credits, generate an API key, dispatch a task to a standard agent, and retrieve the results — all through the API.

This quickstart uses the simplest path: an autonomous task that runs to completion. For a chat-style flow, create a conversation with POST /api/v1/agents/:agentId/conversations; it uses the same task model, but also returns the first message ID for the initial turn.

1. Create an account

Head to the dashboard and sign up. New accounts start with $5 in free credits, so you can test your first runs without a card.

2. Load credits

Open the Billing page in the dashboard when you need more balance and add at least $5 via Polar. New accounts already start with $5 in free credits, and AgentContainer reserves $0.50 at the start of each task, so you need at least that available before you can dispatch anything. Credits cover container time only — AI token costs are billed directly by your provider via your own API key.

3. Generate an API key

Go to Settings in the dashboard and create a new API key. Keys are prefixed with ac_live_ and shown only once, so copy it somewhere safe. You will pass this key in the Authorization header of every request. See the Authentication docs for details.

4. Dispatch your first task

AgentContainer ships with standard agents so you can dispatch tasks immediately — no container or agent setup required. Use the agt_standard_coding agent to run a coding task:

curl -X POST https://agentcontainer.co/api/v1/agents/agt_standard_coding/tasks \
  -H "Authorization: Bearer ac_live_xxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "instructions": "Create a Python script that fetches the top 10 Hacker News stories and saves them to a JSON file."
  }'

The response includes the task id and an initial status of pending. The platform reserves $0.50 from your balance and spins up an isolated container for the agent.

{
  "id": "tsk_abc123",
  "status": "pending",
  "agentId": "agt_standard_coding",
  "createdAt": "2026-03-26T12:00:00Z"
}

5. Poll for results

Poll the task endpoint until status changes to finished (or failed / cancelled). A typical coding task takes one to three minutes.

curl https://agentcontainer.co/api/v1/tasks/tsk_abc123 \
  -H "Authorization: Bearer ac_live_xxxxxxxxxxxx"

When the task finishes, the response includes a summary of what the agent did, the final cost breakdown, and the number of output files.

6. Download output files

First, list the files the agent produced. The response is paginated and returns the files under data:

curl https://agentcontainer.co/api/v1/tasks/tsk_abc123/files \
  -H "Authorization: Bearer ac_live_xxxxxxxxxxxx"

Then download a specific file by its ID:

curl https://agentcontainer.co/api/v1/tasks/tsk_abc123/files/file_xyz789 \
  -H "Authorization: Bearer ac_live_xxxxxxxxxxxx" \
  --output hn_stories.json

Next steps

You have dispatched a task, polled for completion, and downloaded the output. From here you can:

• Read the Concepts page to understand agents, containers, and task lifecycles in depth.
• Learn about Authentication and key rotation best practices.
• Explore the full API Reference for advanced options like conversations, file uploads, and per-task overrides.