Skip to main content

Artefacts & Activity

fibe.gg introduces powerful new ways to interact with and monitor your AI Agents during their operation, ensuring transparency and providing seamless file management.

Agent Activity​

The Agent Activity panel gives you a real-time window into your agent's operation. When an agent is attached to a Playground or running in a standalone chat session, you can monitor its process transparently.

Key Features​

  • Online Activity Timeline: View a continuous chronological timeline of the agent's actions, commands run, and responses.
  • Thinking Chain Visualization: See exactly how the agent is reasoning before it takes action. The UI displays the agent's internal "thinking" processes (e.g., specific to advanced models that expose their reasoning).
  • Grouped Commands: If an agent runs multiple commands consecutively (3+), they are grouped automatically into a collapsible block to keep the chat interface clean and readable.
  • Activity History: Access and download full activity history.
  • Persistent Sidebar: The activity panel remains persistently available on desktop views, allowing you to debug and monitor the agent without leaving your workspace.

Agent Mutters​

Mutters represent the agent's internal "thinking" or reasoning process. While these are often technical or analytical, they are exposed to you for maximum transparency.

  • Real-time Streaming: Watch the agent's thoughts as they are generated.
  • Toggle Visibility: You can collapse or expand mutters to focus on the final actions or the reasoning behind them.
  • Feedback Integration: You can provide feedback specifically on a mutter (see below).

Agent Feedback & Comments​

You can interact directly with your agent's output and reasoning using the Feedback System.

  • Text Selection: Highlight any text within an Artefact or a Mutter to leave a comment or suggestion.
  • Contextual Guidance: Your feedback is injected directly into the agent's context, helping it correct course or refine its approach based on your specific needs.
  • Visual Indicators: Integrated markers show where feedback has been provided in the timeline.

Agent Artefacts​

Agent Artefacts allow you to seamlessly transfer files to and from your AI agents using a dedicated, secure S3-backed storage bucket.

Managing Artefacts​

  • Drag-and-Drop: Easily upload files to the agent's environment by dragging them directly into the chat interface.
  • Multi-type Uploads: Support for multiple file types, automatically categorized with relevant attachment icons.
  • Dynamic File Explorer: View an organized file tree of all the artefacts the agent has generated, downloaded, or that you have provided.
  • Storage: Artefacts can be up to 10MB each and are safely isolated within the agent's running context.
  • Playground Association: Each artefact, feedback, and mutter can be associated with a specific playground_id, indicating which workspace it originated from. Use ?playground_id=ID when listing via the API to filter by workspace.

Logs and Debugging​

For deeper technical inspection, fibe.gg provides comprehensive Agent Logs and Audit Logs.

  • Agent Logs: Display real-time standard output and standard error from the agent sidecar container.
  • Playground Audit Logs: Track all major lifecycle events for the environment to keep an accountable history of creations, updates, and Agent interactions.