Menu
Route plain-text user messages while a /kon:begin session is open. Reuse the same session JSON — never init a second session for sub-turns.
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
Multi-agent design argument — Azusa challenges Mugi's plan, Mugi revises. Used by /kon:design after exploration and initial plan draft.
Multi-agent software engineering workflow. Use this skill whenever the user types a /kon: command — including /kon:team, /kon:design, /kon:quick, /kon:debug, /kon:ask, /kon:review, /kon:review-pr, /kon:hunt, /kon:gc, /kon:research, /kon:describe-issue, /kon:begin, /kon:finish, /kon:summarize, /kon:retro, /kon:address-comments, /kon:todo, /kon:understand-codebase. Orchestrates specialized agents: Azusa (explorer), Jun (researcher), Mugi (planner), Yui (implementer), Mio (reviewer), Sawako (cleaner), Nodoka (summarizer).
This skill should be used by all kon agents at startup, before beginning work, to load relevant memory from public and repo indexes with relevance-based ordering and a 10-entry cap. Consumers: Azusa, Mugi, Mio, Jun, and any future memory-reading agent.
Convention for per-repo skill directories at ~/.kon/projects/<repo-name>/skills/<name>/SKILL.md. Each named subdirectory is one skill. Agents load all SKILL.md files as the first step before memory entries.
When requirements, code behavior, root cause, or review scope is unclear — ask the user before proceeding. Never invent facts, assumptions, or fixes. Applies to every kon stage and every agent.
Keep orchestrator chat lean — artifacts hold full agent output; orchestrator routes by file pointer only. Applies to all /kon:* commands that spawn Task subagents.
| name | interactive-session |
| description | Route plain-text user messages while a /kon:begin session is open. Reuse the same session JSON — never init a second session for sub-turns. |
Owner: orchestrator
Consumers: /kon:begin — all follow-up messages without a /kon: prefix
Core principles (always): follow skills/core-principles — first principles (don't hide the issue); simplest, most concise correct solution.
python3 $KON_ROOT/scripts/kon_session.py active
# prints session id, or nothing
Active = most recent command: "/kon:begin" with status in in_progress | waiting for this project.
If active prints an id → user is in interactive mode unless they sent an explicit /kon:* command.
Never call init for sub-turns inside a begin session.
Append work to the existing session:
python3 $KON_ROOT/scripts/kon_session.py log-turn \
--id <begin-sid> --agent User --summary "<one line paraphrase>"
python3 $KON_ROOT/scripts/kon_session.py complete-agent \
--id <begin-sid> --agent Azusa --summary "<one sentence>"
Begin sessions stay in_progress after each agent — only /kon:finish closes them.
| Signal | Route | Agent(s) |
|---|---|---|
| Question about our code / "where is…" / "how does X work" | ask | 🎸 Azusa read-only |
| External docs / API / "look up…" | research | 📚 Jun |
| Review diff / "check my changes" | review | 📝 Mio |
| Review PR / holistic PR check / "review this PR" | review-pr | 📝 Mio |
| Summarize issue / GitHub issue thread | describe-issue | 📚 Jun |
| Bug / failing test / regression / "X is broken" | debug | 🎸 Azusa → 🍰 Mugi → User → 🎶 Yui → 🧹 Sawako → 📝 Mio |
| Small fix, typo, one file | quick | 🎶 Yui → 📝 Mio |
| Feature, multi-file, unclear scope | team | Azusa → pre-plan gate → Mugi → … |
| Design first, no code yet | design | Azusa → pre-plan gate → Mugi → debate |
| Todo / "remind me to…" / add to list | todo | run kon_todo.py add (no agent) |
| Greeting / meta / "what can you do" | orchestrator | no agent spawn — reply + log-turn User |
When unsure between go and quick, ask once in prose — do not widget.
If the begin session was started with --yolo, append --yolo to any routed pipeline (go/team/design/quick/debug).
/kon:* during beginAllowed — route in-place on the same begin session id. /kon:finish closes the session.
Never call init while kon_session.py active prints an id (hook and CLI both refuse).
Use log-turn --agent User for the command, then run the workflow on that session id.
Hooks auto-record /kon:begin turns — no manual CLI needed for ordinary chat:
log_begin_prompt.py (beforeSubmitPrompt) → User line from each messagelog_begin_response.py (afterAgentResponse) → Assistant line from each replyon_subagent_stop.py → named agent line when a Task subagent finishesOrchestrators may still call complete-agent for explicit pipeline steps; hooks dedupe identical back-to-back entries.
User message paraphrase → optional log-turn --agent User (hook already logs the raw prompt).
/kon:finish after each sub-turn