Feature request: native Hermes approval cards in Chat #857

Problem

When Hermes displays a native tool approval, I have to leave Chat and use Terminal to inspect and answer it. This is especially awkward on a phone. A transcript tool call is not the permission prompt itself, and Chat should not infer approval from it.

Proposal

Would you accept native Hermes approval cards through the existing Chat prompt-card flow?

• Show the native action, warning and available choices without hiding security-relevant details.
• Answer only the choice the user explicitly selects, using native navigation and confirmation.
• Refuse stale, ambiguous, truncated or unreadable panels. Unsupported layouts stay in Terminal; a generic fallback must not silently grant permission.
• Keep the initial scope to tool approvals. No automatic approval, new setting, agent-specific endpoint or persistent approval workflow is requested.

Classic CLI and Ink layouts differ; the supported native versions/layouts should be agreed before implementation. Ordinary clarification questions, model/profile menus, generic command confirmations and Stop behavior are outside this request.

This is a proposal for scope approval before further implementation. Earlier work in #859 is background, not a required implementation. Stop and input-draft protection were addressed separately in #1148 and #1149.

Where

Chat and prompt cards

Where it exists today

Hermes presents approval details and choices in its native terminal UI. This web UI already has approval-card flows for Claude Code and Codex. The request is for the same explicit inspect-and-confirm interaction for Hermes, with its native prompt remaining authoritative.

Workarounds and alternatives

Switch to Terminal, read the complete native prompt and answer there. If a native layout cannot be recognized safely and maintained, continuing to require Terminal is preferable to a guessed or incomplete approval card.

Related requests

• Feature request: Hermes command and individual skill suggestions in Chat #867: command and individual-skill suggestions; approval support would be relevant only to commands that actually require it.
• Feature request: show Hermes informational command results in Chat #1172: informational command results and native pager controls, separate from approvals.
• Feature request: make Hermes skill invocation messages readable in Chat #1173: presentation-only skill invocation messages, independent of permissions.