Oxagen docs
Set agent identity, authority, budget, tools, and skills, then inspect recorded work.
Give each agent an identity and a mandate: its authority, budget, tools, skills, and rules. Oxagen is workforce management for autonomous agents. It checks actions routed through the control plane and keeps the record. Your harness runs the agent.
Start with one agent
Create an organization and workspace, register an agent, and enroll the machine that runs it. Open Runs to inspect the recorded work. Wrap an agent explains what hooks record and where enforcement applies.
Enrollment runs on that machine through the oxagen command, and the Oxagen app installs it:
The newest Oxagen app for each platform
Every version, with SHA-256 checksumsmacOS 12 or newer. Open the .dmg and drag Oxagen to Applications. Builds are not yet notarized, so macOS refuses the first launch. Choose Done, then click Open Anyway in System Settings > Privacy & Security.
- Apple silicon (.dmg)Oxagen_aarch64.dmg
- Intel (.dmg)Oxagen_x64.dmg
Windows 10 or newer, x64.
- Installer (.exe)Oxagen_x64-setup.exe
- Installer (.msi)Oxagen_x64_en-US.msi
x86_64. Install the package for your distribution.
- .deb (Debian, Ubuntu)Oxagen_amd64.deb
- .rpm (Fedora, RHEL)Oxagen.x86_64.rpm
- AppImage (any distribution)Oxagen_amd64.AppImage
On first launch the app links the oxagen and tacho commands onto your PATH, so a new terminal can run them.
On macOS, the app installs new versions on its own and runs them after your next restart. Windows and Linux ask first. How updates work and how to turn them off.
Set its authority and equipment
Roles determine which capabilities a principal may call. Agent roles bound the agent's requests. Use the mandate, Tools, Steering, and Skills pages to manage its equipment and instructions.
For mediated connections, Oxagen holds the downstream credential and uses it on the agent's behalf. The agent's own authentication is separate. Observe mode records what the policy would decide on a governed call and lets the call go ahead. Operator controls, and the budget and model limits on model calls routed through Oxagen, still apply.
Inspect cost and evidence
Spend budgets put period limits on governed consumption. Billing explains credits and model funding. Audit logging describes the security record and its limits.
The knowledge graph supplies scoped context. Agents must request that context through the supported capabilities. Agent memory collects what your agents learn, and a person promotes the useful memories into steering records.
Choose an interface
- CLI: inspect runs, budgets, memory, environments, and secrets.
- REST API: call versioned capability routes.
- MCP: expose the capabilities declared for MCP to a client.
- Architecture atlas: inspect the generated engineering map.
A contract declares which surfaces expose a capability. Those surfaces share the invoke() boundary for authorization, billing, entitlement, and audit. See Capabilities for the contract and route conventions.