Context Panel & Glass Laboratory
The real-time context panel for browsing agents, workflows, tools, scripts, and monitoring execution progress.
Context Panel & Glass Laboratory
The context panel is the collapsible right sidebar in the Zaru chat interface. It serves two purposes: browsing platform resources (agents, workflows, tools, scripts) and monitoring execution progress in real time via the Glass Laboratory.
Opening the Context Panel
Click the sidebar icon in the top-right corner of the chat interface to expand or collapse the panel. On mobile, it opens as a sheet overlay.
Mode-Aware Tabs
The context panel dynamically shows different tabs based on your conversation mode:
| Mode | Tabs |
|---|---|
| Chat | No context panel (conversation only) |
| Agentic | Agents, Tools |
| Execute | Workflows, Tools |
| Workflow | Agents, Workflows, Tools |
| Live | Scripts, Tools |
| Operator | Agents, Workflows, Tools |
Agent Browser
Browse and search deployed agents. Each agent card shows name, description, version, and labels. Click an agent to view details or interact with it:
- Execute opens an inline parameter form and runs the agent directly, showing the result inside the panel.
- Send to Chat injects a structured prompt into the chat input so Zaru can assist with configuration before running the agent.
Workflow Browser
Browse and search registered workflows. View workflow metadata, state definitions, and trigger execution from the panel. Each workflow card offers the same Execute and Send to Chat actions as agents.
Tool Browser
Browse available SEAL tools. Search by name or capability. View tool descriptions and input schemas. Each tool supports direct execution or chat-assisted invocation.
Script Browser
Available in Live mode. Browse saved TypeScript scripts. Search by name or description. From the panel you can:
- Run a script directly
- Insert into chat for modification before running
- Fork a script to create an editable copy
Script persistence is coming soon — saved scripts will appear here automatically.
Glass Laboratory
When an agent or workflow execution is running, the Glass Laboratory renders each step of the execution pipeline in real time:
- Iteration progress — current iteration number and total allowed
- Tool calls — invocations with arguments and results
- LLM reasoning and output — the agent's generated content at each step
- Validation scores and judge verdicts — feedback from iteration judges
- File operations and workspace changes — modifications made during execution
- Workflow state transitions — FSM state changes as the workflow progresses
- Policy decisions — security decisions made by the SEAL Gateway
The Glass Lab updates live via server-sent events. Completed executions remain viewable until the conversation moves on.