> ## Documentation Index
> Fetch the complete documentation index at: https://olyrosstudios.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Olyros Code
> Run AI agents securely where your code and data live — managed from one console, governed and metered end to end.
## AI agents your organization can actually deploy
Every organization is being told to adopt AI agents. Then the practical questions arrive.
Where do agents run? What can they touch? Who approved that credential? What did last
month's agent activity cost, and on what?
Most agent platforms answer by asking you to send your code and data to their cloud, and
to take visibility on faith. Olyros Code takes the opposite approach.
### Manage centrally. Run on infrastructure you control.
You define agents, environments, credentials, and schedules in one cloud console. The work
itself executes on infrastructure you control — your own servers or laptops via the Olyros
orchestrator, or inside your own cloud account, with Google Cloud and AWS available today
and Azure coming soon.
Agents run in isolated containers next to your real repositories and data, behind your
firewall, with no inbound network configuration. If connectivity drops, work continues and
syncs when it returns. Your code never has to leave your environment.
### Agents that join your team, not another tool to babysit
Work reaches agents the way it reaches people. Label a GitHub issue or Jira ticket with an
agent's name and it becomes that agent's task: a scoped checkout of the right repositories,
the work done in an isolated container, a pull request opened under the agent's own git
identity, and results posted back on the ticket.
Recurring schedules can file tickets on the board your team already uses, so agent work
shows up in the same sprint reports as everyone else's. And agents are available in Slack —
which non-technical teams especially love: a capable agent with real skills and a safe
sandbox, in the channel they already work in, with nothing to install.
Slack is also where agent capability spreads beyond engineering. The people closest to the
work run skills directly in Slack — no scripts passed around, and no credentials ever in
their hands. Skills are written once, by an engineer or by anyone willing to learn to work
with an agent, saved to GitHub and versioned like any code, then exposed by admins through
Olyros Code workflows as simple tools. Your organization shares capability, not credentials.
And once a tool has earned trust, it can run on a schedule — reports, analyses, and audits
that happen on their own, with every run metered.
### Governance you can show your security team
Every interaction is metered — token and dollar cost per agent, model, integration,
and channel, with trends and CSV export.
Agents receive short-lived, minimally scoped credentials that are never written to
disk — for example, one-hour GitHub tokens limited to a single task's repositories.
Multi-tenant isolation is enforced at multiple layers and continuously validated by
an automated security test suite.
Role-based access, enforced single sign-on, audit logging, and admin approval before
any machine can receive work.
Read more: [How it works](/product/how-it-works) · [Security & trust](/product/security) · [Integrations](/product/integrations)
***
## Get started
Signup is open — connect a repository and dispatch your first agent task today, or bring
us a workflow and we'll walk you through what it looks like on the platform.