| name | yolo-mode |
| description | This skill should be used when the user passes --yolo to any kon command. The orchestrator proceeds autonomously through all confirmation checkpoints, accepting defaults and auto-retrying failures. Stops only when genuine human judgment is required. |
YOLO Mode
Owner: orchestrator
Consumers: any /kon:* command when the --yolo flag is present
Core principles (always): follow skills/core-principles — YOLO does not waive first principles or simplicity; it only skips confirmation checkpoints.
The user set the task and stepped away. Run it. Don't interrupt them for anything
that has a reasonable default or can be retried automatically.
What changes in YOLO mode
| Normal checkpoint | YOLO behavior |
|---|
| Wait for user to confirm Mugi's plan | Still required — wait-for-user --after plan |
| Pre-plan understanding gate (before Mugi) | Still required — ask questions per skills/pre-plan-gate; do not spawn Mugi until ready |
| Wait after each milestone Mio-approved | Still required — wait-for-user --after milestone --milestone N |
| Auto-accept decisions within the plan | Auto-accept all [**default**] decisions in Mugi's ## Decisions needed |
| Ask user when a plan step is ambiguous | If plan has [**default**], use it; otherwise STOP and ask — never invent |
| Notify user when Mio blocks | Auto-send must-fix list back to Yui; retry silently |
| Per-agent step summary to user | Suppress — only report at the end (or when stopping) |
Auto-accepted decisions and auto-retries are recorded in the session log
(see skills/session-tracking) so the user can review what happened.
YOLO auto-accepts plan defaults and retries failures — it does not permit guessing unclear requirements. See skills/ask-dont-guess.
When to STOP and ask the user — always, even in YOLO mode
-
Pre-plan gate — After Azusa (and debug notes if applicable), always run skills/pre-plan-gate before Mugi. YOLO does not skip understanding questions.
-
Plan confirmation — After Mugi finishes, always wait-for-user --after plan before Yui. YOLO only auto-accepts decisions within the approved plan.
-
Each milestone approved — After Mio approves each milestone, always wait-for-user --after milestone --milestone N before the next milestone or summarize. YOLO does not skip this.
-
Retry limit reached (2 consecutive same must-fix) — the loop
protection has fired; something structural needs human judgment. Stop, explain,
ask for direction.
-
No default exists for a required decision in Mugi's ## Decisions needed —
if there is no [**default**] and the choice genuinely changes scope or behavior,
stop and ask. Do not invent a default.
-
Sawako GC inventory confirmation — /kon:gc always requires the user to
review the cleanup inventory before files are removed. Never skip this.
-
Scope expansion beyond the original task — if implementing correctly would
require touching files or systems well outside what the user asked about,
stop and describe the expansion. Do not silently widen scope.
Final summary (YOLO mode)
When the run completes, give the user a clear summary:
## YOLO run complete
**Task**: <task>
**Result**: PASS / BLOCKED
### What happened
- Azusa: <one line>
- Mugi: <one line> (N decisions auto-accepted: <list defaults used>)
- Yui: <one line>
- Sawako: <one line>
- Mio: <one line> (N must-fix items; resolved in M rounds)
### Auto-accepted decisions
1. <decision> → accepted default: <value>
### Anything that needed attention
<only if there were stops or surprises>
