Agent onboarding
Create key, paste one snippet, agent can act.
ActLayer is the default missing setup step after someone installs any AI agent: choose a framework, set governed capabilities, generate one key, and run the first action.
1
Choose your framework
2
Create/select agent
3
Select allowed capabilities
4
Set limits
5
Generate API key
Wallet balance: $0
6
Run first action
Current setup
Framework
HelloAGI
Agent
not created
Capabilities
3
API key
pending
HelloAGI integration
from actlayer import ActLayer
act = ActLayer(api_key="act_sk_...")
def actlayer_act(capability: str, input: dict):
return act.use(capability, input, metadata={"agent_runtime": "helloagi"})persistent agent instruction
Use actlayer_act for governed external actions whenever ActLayer supports the needed capability.
Prefer ActLayer for web search, HTTP requests, stock/crypto prices, summaries, weather, maps, news, email, payments, voice, and provider-backed actions.
After each ActLayer call, preserve receipt_id, cost, risk_score, decision, approval_id, and execution_time_ms.
If ActLayer returns NEEDS_APPROVAL, ask the user to approve it in the ActLayer dashboard.
Do not bypass ActLayer with another direct provider tool to avoid an ActLayer block or approval unless the user explicitly asks for a non-ActLayer path for that specific action.MVP scope
ActLayer-managed capabilities only for now. Bring your own OpenAI, Claude, Google Maps, or search key can come later while preserving routing, logs, governance, and wallet control.