Project Settings Overview

Project Settings control how a specific project is configured, executed, and connected inside Origin. While account-level settings manage identity and authentication, this area defines how the project behaves — how sandboxes run, which models are available, what secrets agents can access, and how usage is tracked.

These settings apply to all activity inside the project, including task execution, agent sessions, pull request creation, and sandbox-based runs. They do not directly modify your repository. Instead, they define the operational rules under which agents access code, execute changes, and use resources.

The left sidebar for a project contains:

• Project Details: project name, description, team, milestones, cycles, and metrics
• Project Context: context documents and instructions shared across the project
• Sandboxes: live sandbox instances and their resource status
• Models: default code model and model category access controls
• Secrets: environment variables and API keys used during execution
• Logging Integrations: external log monitoring connections
• Project Usage: compute costs, token consumption, and activity metrics

Each section is described briefly below. Follow the links for full details on each area.

Project Details

The Project Details section defines the identity and metadata of the project.

Here you can update the project name, description, and team assignment. Changes are saved using the Save Changes button at the top right.

The section also includes:

• Milestones: named checkpoints for tracking project progress, each with a target date. Add milestones using the + Add button.
• Cycles: time-boxed iterations (for example, one-week sprints) with status badges showing whether a cycle is Active or Upcoming.
• Project Metrics: a read-only block showing the masked API Key (with a copy button), Project ID, and total task count.
• Additional Information: the linked repository, active branch, and project creation date.

At the bottom, the Archive Project option hides the project from the main view while preserving all data and tasks. Archived projects can be restored later.

→ Learn more about Project Details

Project Context

The Project Context section lets you attach persistent instructions and reference documents that are shared across the project.

Context added here is available to agents during sessions and task execution, without needing to be re-attached each time.

→ Learn more about Project Context

Sandboxes

The Sandboxes section shows the workspace history for the project — all sandbox instances that have been created, along with their current status and resource allocation.

Each sandbox entry displays:

• a generated instance identifier (for example, trial-96acac35)
• status badge: Running or Stopped
• TDX Sandbox badge, confirming the instance runs on TDX-backed confidential infrastructure
• allocated resources: CPU cores, memory, and disk (for example, 4 cores / 8 GB / 10 GB)
• a unique instance UUID

A cost summary is shown at the top right, reflecting the total spend across active workspaces in the project.

→ Learn more about Sandboxes

Models

The Models section has two parts: selecting the default code model for the project, and controlling which model categories are available.

Code Model lets you choose the primary model used by the project assistant. This becomes the default across chats and tasks unless overridden in a specific session.

Model Access controls which categories of models are available within the project. Two categories can be toggled independently:

• Allow ZDR Models: enables standard zero-data-retention models, including Vercel AI Gateway models and non-TEE OpenCode providers. Suitable for routine development work.
• Allow TEE Models: enables attested TEE models, including NEAR, Phala, and o.llm-backed models. These run on confidential infrastructure and are appropriate for sensitive inference.

At least one category must remain enabled at all times. Changes can be applied immediately using the Instant Apply button.

→ Learn more about Models

Secrets

The Secrets section manages environment variables and API keys injected into the execution environment during agent runs.

You can add individual secrets or import them from a .env file. Secrets are encrypted and masked after saving — their values cannot be retrieved once stored.

Secrets are never written to the repository, committed to Git, or exposed in diffs. They are only available to the execution environment during sandbox runs.

→ Learn more about Secrets

Logging Integrations

The Logging Integrations section connects external log monitoring services to the project.

Currently, Origin supports connecting BetterStack. Once connected, BetterStack monitors your log sources, tracks incidents, and can automatically create tasks from log anomalies.

To connect, paste your BetterStack API token and click Connect. The token is stored securely and used only for API access to your BetterStack telemetry data.

Once connected, the integration dashboard shows four summary figures:

• Log Sources: the number of BetterStack log sources available to scan
• Tickets Created: tasks generated from log analysis
• Tickets Resolved: resolved tickets originating from log anomalies
• Automatic Scans: a configurable scan frequency (for example, every 24 hours) with an on/off toggle

You can also run a diagnostic manually using the Run Diagnostics panel. Select a log source, a time window, and a scan mode (for example, "All logs"), then trigger the scan. The diagnostic runs through five stages: Connecting, Fetching Logs, Analyzing Logs, Consolidating, and Creating Tasks. A progress bar tracks the current stage.

→ Learn more about Logging Integrations

Project Usage

The Project Usage section shows resource consumption and activity metrics for the project.

The top of the page shows four summary figures:

• Compute Cost: total spend on sandboxes and workspace runs
• Total Tokens: total tokens consumed across model calls in this project
• LLM Cost: spend attributed to AI model usage
• Grand Total: combined cost across all services

The Sandbox Costs table lists each sandbox instance with its status, resource allocation (CPU / memory / disk), and trial count. The per-session billing rate and billable session count are shown alongside the billed total.

The AI Model Usage section breaks down token consumption by model once data is available.

The AI Line Edits section shows a heatmap of code edit activity over time, similar to a contribution graph. It can be filtered to show all activity, LLM-driven edits, or task-driven edits. Summary stats include most active month, most active day, and current and longest streaks.

A Usage Leaderboard is also available further down the page.

→ Learn more about Project Usage

How Project Settings Are Typically Used

Teams usually visit Project Settings when:

• updating the project name, description, or team assignment
• reviewing or adding milestones and cycles for sprint planning
• checking sandbox status and resource allocation
• selecting a default model or adjusting model category access
• adding secrets before running tasks that call external services
• connecting a log monitoring integration
• reviewing compute and token costs for the project

In short, this area defines the operational environment of the project — repository access, compute resources, model configuration, and security boundaries — while leaving the source code itself unchanged until tasks explicitly generate updates.