Show selected hosted-agent configuration in the workflow overview

[jaeyong8687]

Description

The build configuration rows do not expose what has been selected, so users cannot verify the agent’s model, toolboxes, skills, or guardrails from the workflow overview.

Each row should summarize its current state and support at least: loading, empty, configured, unavailable, and error. The design follow-up should define the compact summary; engineering should then connect it to authoritative configuration state.

Expected outcome

• Model shows the selected deployment or an explicit empty state.
• Toolboxes and skills show selected names or meaningful counts.
• Guardrails show enabled, disabled, unavailable, or error state.
• Loading and failure states do not look like valid empty configuration.
• The summary remains visible when the configuration section is collapsed.

Evidence — C63

Problem: Unable to know what's the inside of this config

Suggestion: Better to show what are the selected models, toolboxes, skills and guardrails.

Relationship

Parent: #602

Priority

P2 — Missing configuration transparency increases uncertainty and review errors but does not independently block creation.

Size

M — Requires compact-state design plus integration with model, toolbox, skill, and guardrail state.

Source

Internal UX audit, July 27, 2026. FigJam evidence. POC report: vscode/extension/ai-mlstudio/docs/experience-audit/agent-canvas-2026-07/report.md (not yet published upstream).

[microsoft-github-policy-service]

Thank you for contacting us! Any issue or feedback from you is quite important to us. We will do our best to fully respond to your issue as soon as possible. Sometimes additional investigations may be needed, we will usually get back to you within 2 days by adding comments to this issue. Please stay tuned.