gistfile1.txt

The sidebar appears to contain:

| Option                 | What it does                                                           |
| ---------------------- | ---------------------------------------------------------------------- |
| **General**            | Organization-wide settings, branding, permissions, and defaults.       |
| **Connections**        | Configure integrations with external services, APIs, and tools.        |
| **Sessions**           | View and manage Devin's active and past working sessions.              |
| **Devin**              | Core Devin settings and behavior configuration.                        |
| **DeepWiki**           | Generate and browse AI-created documentation for repositories.         |
| **Schedules**          | Create recurring or scheduled Devin tasks.                             |
| **Skills & Rules**     | Define custom instructions, workflows, and behavioral rules for Devin. |
| **Environment**        | Configure runtime environments, containers, and execution settings.    |
| **Knowledge**          | Store documentation and reference material Devin can use.              |
| **Playbook Knowledge** | Structured organizational procedures and operating guides.             |
| **Secrets**            | Manage API keys, credentials, and sensitive configuration values.      |
| **Repositories**       | Connect and manage source code repositories.                           |
| **Membership**         | Manage users, seats, roles, and billing membership.                    |
| **Devin API**          | Configure and access Devin's API endpoints and credentials.            |
| **Analytics**          | Usage metrics, activity reporting, and performance insights.           |

The overall structure is interesting because it suggests Devin separates knowledge into three layers:

1. **Knowledge** → raw documents and information.
2. **Playbook Knowledge** → documented procedures ("how we do things").
3. **Skills & Rules** → instructions that directly affect agent behavior.

That's a fairly mature enterprise-agent architecture: **data → process → behavior**.