Developers

Two languages, both first-class.

Python and TypeScript, kept at feature parity and released together. Two rather than eight, because a half-maintained SDK in your language is worse than none — you find out which parts were neglected at the point you depend on them.

Read the docs Quickstart

Your logic stays yours

The SDK gives you identity, tool calls, permissions, state and tracing. How your agent decides anything is your code, and the platform does not impose a graph, a chain or a framework.

Tools are typed from the registry

A granted capability arrives with its declared input and output types. Calling one with the wrong shape fails at your type check rather than at runtime in production.

Permission failures are explicit

Calling a capability the agent does not hold raises a specific, catchable error naming the capability and the grant it would need. Not a generic 403.

The same manifest, either language

An agent's manifest is language-independent. A Python agent and a TypeScript agent with identical manifests have identical permissions, limits and audit behaviour.

Roughly what it looks like

Illustrative and deliberately short. The reference is at docs.agentlabora.com, and pretending a marketing page is documentation wastes an afternoon.

An agent is a function with a manifest

python

from labora import Agent, tool

agent = Agent(
    name="invoice-summary",
    tools=["invoice.get", "customer.get"],
)

@agent.run
async def run(ctx):
    inv = await ctx.call("invoice.get", ref=ctx.input["invoice"])
    return {"overdue_days": inv.overdue_days}

Where it stops

The SDK cannot widen what an agent may do. Granting a capability is a manifest and review change, not something you can do from code at runtime — which is occasionally inconvenient and is the point. There is also no local escape from the permission layer: an agent running on your laptop against production capabilities is still checked.

Developer platform

The rest of it.

QuickstartA deployed agent in under fifteen minutes.APIAgents, runs, tools and events over HTTP.CLICreate, test, deploy, tail logs, roll back.Custom toolsExpose your own service as a permissioned capability.Agent manifestThe declarative file your agent lives in.SandboxSynthetic data, mock tools, no production reach.EvaluationsTest suites that run in your pipeline.PublishingShip an agent to the marketplace.DocumentationThe reference, the guides and the API explorer.

Give AI a job.

Build an agent, choose what it can access, define what it may do, and deploy it to work safely across your business.

Build an agent Build with the API

No code required · Full SDKs and API · Stop any agent instantly · [email protected]