Ódinn Forge documentation

Ódinn Forge brings conversations, projects, goals, memory, tools, scheduled work, and an inspectable activity history into one local console. It runs on a machine you control and connects to the model provider—or local model server—you choose.

The current release is v0.4.0-beta.3. It is a real public beta, not a stable release. The core local operator path is substantially verified, while several real-world validation campaigns remain open. Start with Public beta status if you need the exact boundary.

Find your path

I want to try Ódinn

Follow Install, complete Onboarding, and send Your first conversation.

I already use Ódinn

Explore the operator console, memory, models and providers, tasks and activity, and automatic improvements.

I need to operate it safely

Read Privacy, the security model, known limits, and state backup and restore.

I want the technical details

Use the CLI reference, gateway API, configuration reference, event ledger, and architecture overview.

Product boundaries at a glance

• Verified in the public beta: the single-operator local workflow, loopback gateway, console, audited runtime, durable local records, release archives, checksums, and rollback paths.
• Provider- or platform-dependent: live model services, local model servers, websites, browser engines, authentication flows, and operating-system behavior.
• Experimental and off by default: the seven Labs capabilities.
• Enabled by default and bounded: Automatic improvements, which can apply only predefined reliability tuning and keeps a restore point.
• Unsupported: hostile-code containment, public exposure of the single-user gateway, and guaranteed replay or rollback of arbitrary external effects.

Ódinn has no built-in product telemetry. A cloud model provider still receives the prompts, recalled context, and tool results sent to its API. Read Privacy before connecting accounts or providers.