Administration & Management — User Guide

Audience: Workspace admins and users who need to monitor job history, track AI usage costs, configure runtime environments, and manage team settings.

Overview

Workspace is the primary boundary for metadata: connections, repositories, jobs, Agent Intelligence, Knowledge Base, billing hooks, and API keys. APIs scope data by workspace — always confirm the active workspace in the selector before running pipelines or sharing secrets. Invited members share the same workspace subject to RBAC; see Team Settings below.

Dagen provides five administration views:

Workspace sharing and isolation

• A workspace is the tenant boundary for metadata: database connections, repos, jobs, Agent Intelligence, Knowledge Base, billing hooks, and API keys.
• APIs enforce workspace_id scoping—users cannot query another workspace by ID.
• Share workspace (owners): invite by email, assign role, optional message; invitees use Accept Invite. Members share assets subject to RBAC.
• Always select the correct workspace in the workspace switcher before connecting secrets or starting long jobs.
• Do not upload secrets into the Knowledge Base—treat uploads as retrievable by agents in that workspace.

Team administration lives under Team Settings (/team-settings) for invites and membership.

Where runs surface (cross-feature)

Runtime engine: many failures are environment issues—wrong Runtime Environments configuration (see Part 3 below), expired credentials, network egress, or cluster capacity. Re-run Test on the runtime card; confirm default ingestion runtime in the ingestion header; for Spark validate cluster / region / namespace fields.

Agent-assisted debugging: in AI Chat, attach pipeline, database, or workflow context and paste error excerpts; prefer Guided or Semi execution mode until validated.

Part 1 — Job History (Activity)

The Job History page shows a log of all agent interactions in your workspace.

Browsing Jobs

The main table has these columns:

Use pagination controls at the bottom to navigate through results.

Viewing Job Details

1. Click View Details on any row.
2. The detail dialog shows:
   • Status chip.
   • Start Time and End Time.
   • Final Summary of the job outcome.
3. Below the summary, the Interaction History section lists every step:

Each tool step also shows its duration in milliseconds.

If no steps were recorded, you see: "No detailed steps recorded for this job."

Part 2 — Usage Analytics (Chargeback)

The Usage Analytics page tracks AI model usage, costs, and performance across your workspace.

Filtering

Use the filter bar at the top:

When Custom Range is selected, date pickers for Start Date and End Date appear.

Summary Cards

Four cards at the top provide high-level metrics:

Charts

Usage Timeline

A line chart showing cost and token usage over time. Switch the interval with the selector: Hourly, Daily, Weekly, or Monthly.

Usage by Model

A doughnut chart showing cost distribution across AI models.

Model Usage Statistics Table

Recent Usage Details Table

A paginated table of individual API calls:

Navigate pages with the Previous / Next buttons ("Page X of Y").

Budget Alerts

The Budget Alerts section at the bottom shows any configured cost thresholds that have been triggered, with alert type, message, and timestamp.

Part 3 — Runtime Environments

The Runtime Environments page (/runtime-environments) configures where your compute workloads run.

Runtime Categories

The page organizes runtimes into three categories, each with its own tab:

Switch between All Runtimes and category-specific tabs. Each tab shows a count chip.

Adding a Runtime

1. Click Add Runtime in the header.
2. Select the runtime type from the menu:
   • Data Ingestion Runtime — for ingestion jobs (CDC, batch sync).
   • Spark / Data Processing — for Databricks, Snowflake, Dataproc, etc.
   • Python Execution — for Docker, Kubernetes, Cloud Run, Lambda.
3. Fill in the form:
   • Runtime Name (required).
   • Provider-specific configuration fields (host, cluster, credentials, etc.).
   • Optionally check Set as default runtime for this workspace.
4. Click Create Runtime.

Testing a Runtime

Click Test on any runtime card. A connection test runs and the result appears in a snackbar (success or error with details).

Setting the Default

Click Set Default on a runtime card to make it the workspace default. The default runtime is marked with a Default chip.

Editing and Deleting

• Click Edit to modify a runtime's configuration.
• Click the Delete icon and confirm: "Are you sure you want to delete {name}? This action cannot be undone."

Setup Guide

Click Setup Guide in the header for provider-specific setup instructions. Tabs cover Data Ingestion, Serverless Spark, Databricks, Snowflake, and Google Cloud.

Part 4 — Team Settings (Admin Only)

The Team Settings page (/team-settings) is restricted to workspace admins. Non-admin users see: "Access Restricted — You do not have admin privileges to manage team settings, users, or billing configuration."

Sections

Inviting Users

Use the onboarding section to invite new team members by email. They receive an invitation to join your workspace.

Managing Members

The team members section lets you search for users, view their roles, and update permissions.

Part 5 — Model Settings (Admin Only)

For detailed model configuration, see the dedicated Model Settings page.

Troubleshooting