Free SKILL.md scraped from GitHub. Clone the repo or copy the file directly into your Claude Code skills directory.
npx versuz@latest install auroraphtgrp01-auroraphtgrp-coding-kit-temp-skills-ak-mioflow-mioflow-skills-exploringgit clone https://github.com/auroraphtgrp01/auroraphtgrp-coding-kit.gitcp auroraphtgrp-coding-kit/SKILL.MD ~/.claude/skills/auroraphtgrp01-auroraphtgrp-coding-kit-temp-skills-ak-mioflow-mioflow-skills-exploring/SKILL.md---
name: exploring
description: >-
Use when a feature request needs user-facing decisions captured in
history/<feature>/CONTEXT.md before MioFlow planning. Clarifies fuzzy scope
without implementation research or bead creation.
metadata:
version: '1.0'
ecosystem: mioflow
position: '2 of 9 - after using-mioflow, before planning'
dependencies: []
---
# Exploring Skill
If `.mioflow/onboarding.json` is missing or stale, stop and invoke `mioflow:using-mioflow`.
Exploring turns fuzzy intent into locked decisions in `history/<feature>/CONTEXT.md`.
## Hard Gates
- Ask one question at a time; wait for the user before asking the next.
- Do not answer your own question.
- Do not research implementation, propose architecture, create beads, or write code.
- Do not invoke planning. End by handing the user to `mioflow:planning`.
## Flow
1. **Scope**
- Classify: `Quick`, `Standard`, or `Deep`.
- Read `history/learnings/critical-patterns.md` and `.mioflow/state.json` if present.
- If the request spans independent subsystems, pick one and defer the rest.
2. **Domain**
- Classify each applicable type:
- `SEE`: user-visible surface
- `CALL`: API, CLI, webhook, SDK, service interface
- `RUN`: job, script, service, or pipeline
- `READ`: docs, emails, reports, notifications
- `ORGANIZE`: data model, file layout, taxonomy, config
- Load `references/gray-area-probes.md` and choose only relevant probes.
3. **Gray Areas**
- Generate 2-4 unstated product decisions that would make planning guess.
- Do a quick scout only:
```bash
rg "<feature-keyword>" src app packages --glob "*.{ts,tsx,js,jsx,py,md}" | head -20
```
- Read 2-3 relevant files; cite existing patterns in questions.
- Exclude implementation choices, performance tuning, and new scope.
4. **Socratic Locking**
- Ask one concise question per message, preferably single-choice.
- Start broad, then narrow into constraints.
- After each decision, confirm it and assign a stable ID: `D1`, `D2`, `D3`.
- For scope creep: mark it as deferred and return to the current question.
5. **Context Assembly**
- Write `history/<feature-slug>/CONTEXT.md` from `references/context-template.md`.
- Include boundary, domain types, locked decisions, scout paths, refs, questions, and deferred ideas.
- Use concrete language. No placeholders, TODOs, or vague preferences.
- Spawn one fresh reviewer with no history. Check completeness, contradictions, vague decisions, missing IDs, and blockers. Max two loops.
6. **State And Handoff**
- Update `.mioflow/state.json`:
```json
{
"active_skill": "exploring",
"feature_slug": "<feature>",
"phase": "exploring-complete",
"summary": "Exploring complete. CONTEXT.md is ready for planning.",
"next_action": "Invoke mioflow:planning.",
"focus": "history/<feature>/CONTEXT.md"
}
```
- Tell the user:
```text
Decisions captured. CONTEXT.md written to `history/<feature>/CONTEXT.md`.
CONTEXT.md is the source of truth for downstream agents.
Invoke mioflow:planning to research, propose phases/stories, and wait for approval.
```
Anti-patterns: bundled questions, deep implementation analysis, architecture proposals, beads, code, or skipping decision locking.
References: `references/gray-area-probes.md`, `references/context-template.md`.