kon

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.

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.

Keep orchestrator chat lean — artifacts hold full agent output; orchestrator routes by file pointer only. Applies to all /kon:* commands that spawn Task subagents.

Mandatory understanding check after exploration and before any Mugi planning step. Orchestrator asks high-level and implementation questions; do not spawn Mugi until answers resolve material gaps. Applies to /kon:team, /kon:design, /kon:debug.

This skill should be used by the kon orchestrator at the start of every command and after every agent step, to write a JSON session file to ~/.kon/projects/<repo-name>/sessions/ that the dashboard reads.

This skill should be used when orchestrating the full kon teammate flow — Azusa explores, Mugi plans, Yui implements, Sawako cleans, Mio reviews. Applies to /kon:team.

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.

Compressed chat output mode. Strip filler prose from orchestrator and narrator responses while keeping full technical accuracy. Apply to conversational output only — narration beats, status lines, agent summaries in chat. Never apply to structured artifact files (plan.md, review.md, debug.md) or hook-validated sections. Activate with `/caveman [lite|full|ultra]`. Deactivate with "normal mode".

This skill should be used by the kon orchestrator on every /kon command, to frame the run with Ui narration at the opening, the closing, and stuck-point beats, and to enforce the emoji prefix on every mention of a kon agent or narrator.

This skill should be used by the kon orchestrator on every /kon command, alongside narration, to govern the conversational conduct of the main dialogue — AskUserQuestion widget discipline and word-choice norms.

This skill should be used for /kon:hunt — read-only bug analysis from source code with best-effort repro SQL and test suggestions. Azusa only; no fixes, no repo writes.

This skill should be used when drafting a git commit message from staged or proposed changes. Produces a user-impact subject and a concise body. Deliberately avoids motivation paragraphs (those belong in the PR description, not the commit log).

Kon's two highest-priority principles for design, implementation, and review — first principles (don't hide the issue) and the simplest, most concise correct solution. Applies to every agent, skill, and workflow stage.

When to spawn 📚 Jun for external lookup — WebSearch/docs/APIs — vs 🎸 Azusa for codebase exploration. Used by /kon:research, /kon:team, and /kon:design.

This skill should be used when handling failures in pipeline command flows — Mio blocking with NEEDS_CHANGES / BLOCKED, and infinite-loop protection for repeated same must-fix. Applies to /kon:quick, /kon:team, /kon:debug.

Summarize a GitHub issue and all discussion comments. Used by /kon:describe-issue (Jun). Writes sessions/<id>/issue-summary.md.

This skill should be used by the kon orchestrator when saving memory — from

Task subagents must use the orchestrator's model — Cursor defaults to Composer if model is omitted. Applies whenever spawning Task subagents in kon.

Holistic GitHub PR review — code diff, PR description, existing review comments, and linked issues. Used by /kon:review-pr (Mio). Read-only; no gh pr merge or code edits.

This skill should be used by the orchestrator at the end of pipeline commands (team, quick, design, debug, gc, address-comments) after /kon:summarize, to propose session learnings for public or repo memory with human confirmation before write.

This skill should be used when performing code review on a diff (whether implemented by another agent or an external PR), enforcing a strict reviewer stance, applying a mandatory 7-item golden checklist, demanding evidence, and giving specific改法 instead of vague critique.

Map a codebase for learning — glossary, architecture, flashcards, and quiz. Produces a PDF guide and local HTML study pack.