Quickstart
Deploy your first AI agent from a clean account.
# 1. Create a workspace
Open Console and sign up or sign in. New users will create or select a workspace before reaching the console.
# 2. Install the CLI
▓ shell
npm install -g @parel/cli
# 3. Authenticate
Create a workspace API key in Console > Settings, then:
▓ shell
parel login # API key: pk_your_key_here # Logged in.
# 4. Configure provider access
The example below uses Anthropic for the model and E2B for the sandbox. Either add your Anthropic and E2B keys in Console > Settings, or make sure platform billing is configured for usage paid through PAREL.
# 5. Create agent.yaml
This is the only file you need. It declares your agent's model, tools, memory, and budget.
▓ agent.yaml
version: "1"
agent:
name: my-first-agent
model:
provider: anthropic
model: claude-sonnet-4-20250514
plugins:
- system-static:
prompt: "You are a helpful research assistant."
- sandbox-e2b
- memory-rolling-summary
- budget-cap:
max_usd: 10
runtime:
maxSteps: 100# 6. Deploy
▓ shell
parel deploy agent.yaml # Deployed: my-first-agent # id: agent_a1b2c3d4
# 7. Chat
▓ shell
parel chat --agent agent_a1b2c3d4
PAREL Chat
session: abc123def456
type /quit to exit, /messages for conversation log
$ What's trending on Hacker News?
▸ Let me check...
[tool] bash({"command": "curl ..."})
[result] {"output": "..."}
▸ Here's what I found: ...
$ /quit# 8. Monitor
Open parel.sh/console to see your agent's sessions, costs, and live traces.
Next: Read Concepts to understand how the kernel, plugins, and sessions work together.