Menu
Ask the user questions. Use instead of ask_user (which is disabled).
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Report on currently-running SASE agents. Use when the user asks "what's running?", "agent status", "status report", or any question about live/background agents.
Reference for sase bead commands (create, update, list, search, ready, show, dep). Use when working with beads.
Inspect prior sase agent chat transcripts. Use when the user asks about previous chats, chat transcripts, prior agent conversations, "what did agent X say?", "summarize the previous agent", or any question about an earlier sase agent's prompt or response.
Create an implementation plan. Use instead of plan mode (which is disabled).
Create an implementation plan. Use instead of plan mode (which is disabled).
Create an implementation plan. Use instead of plan mode (which is disabled).
| name | sase_questions |
| description | Ask the user questions. Use instead of ask_user (which is disabled). |
Before doing anything else, run this command to record that you are using this skill:
sase skill use sase_questions --reason "<one-line reason for using this skill>"
Use this skill when you need user input. This replaces OpenCode's native ask_user tool.
sase questions '<json>'
[
{
"question": "Full question text (required)",
"header": "Short sidebar label (optional)",
"options": [{ "label": "Option label (required)", "description": "Details (optional)" }],
"multiSelect": false
}
]
Single question with options:
sase questions '[{"question": "Which database should we use?", "options": [{"label": "PostgreSQL", "description": "Relational, mature"}, {"label": "SQLite", "description": "Embedded, simple"}]}]'
Multiple questions:
sase questions '[{"question": "Approach?", "header": "Approach", "options": [{"label": "A"}, {"label": "B"}]}, {"question": "Include tests?", "options": [{"label": "Yes"}, {"label": "No"}]}]'
On success, sase questions writes a durable handoff marker and sends SIGTERM to the current agent runner process
group. The runner recognizes this as an intentional question handoff, creates a command-backed UserQuestion gate,
yields its runner slot while it waits, and reacquires a slot before continuing. The answer is added to the Q&A history
and reconstructed follow-up prompt; the interrupted provider turn does not return normally.
Do not poll question request or response files. ACE, mobile, and Telegram submit the complete validated form through the same write-once gate command, and the runner observes the terminal response mechanically.