tool_use: add multiplechoice support to Agent
This implements the "dumb" approach - the tool itself just tells
the llm that it rendered the options to the user, and it's done.
If the user selects one of the options, we paste its response text
into the chat input textarea on the frontend. The user is of
course free to ignore the question or the options presented.
This keeps no association between user response and the original
tool_use block that solicited it from the user. I.e. the user
response message doesn't include the original tool_use_id value
it it. It looks as though the user typed it by hand.
diff --git a/termui/termui.go b/termui/termui.go
index d0bd130..ba08c05 100644
--- a/termui/termui.go
+++ b/termui/termui.go
@@ -46,6 +46,11 @@
✏️ {{.input.file_path -}}
{{else if eq .msg.ToolName "codereview" -}}
🐛 Running automated code review, may be slow
+{{else if eq .msg.ToolName "multiplechoice" -}}
+ 📝 {{.input.question}}
+{{ range .input.responseOptions -}}
+ - {{ .caption}}: {{.responseText}}
+{{end -}}
{{else -}}
🛠️ {{ .msg.ToolName}}: {{.msg.ToolInput -}}
{{end -}}