Claude-skill-registry brainstorming-exploring
The EXPLORING phase gathers context about the project and forms an initial understanding
git clone https://github.com/majiayu000/claude-skill-registry
T=$(mktemp -d) && git clone --depth=1 https://github.com/majiayu000/claude-skill-registry "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/data/brainstorming-exploring" ~/.claude/skills/majiayu000-claude-skill-registry-brainstorming-exploring && rm -rf "$T"
skills/data/brainstorming-exploring/SKILL.mdEXPLORING Phase
The EXPLORING phase gathers context about the project and forms an initial understanding of what needs to be built.
Purpose
- Understand the current project state
- Read relevant files and documentation
- Check recent git history for context
- Form an initial list of items/topics to discuss
Step 0: Query Kodex
Before reading files, check project knowledge base for relevant context.
Topic Inference
- Get current work item from collab-state.json
- Extract keywords from item title/description
- Build topic candidates:
{keyword}{keyword}-patterns{keyword}-conventions
Query Process
FOR each candidate topic name: Tool: mcp__plugin_mermaid-collab_mermaid__kodex_query_topic Args: { "project": "<cwd>", "name": "<candidate>" } IF found: Add to context
Example
For work item "Add user authentication":
Tool: mcp__plugin_mermaid-collab_mermaid__kodex_query_topic Args: { "project": "<absolute-path-to-cwd>", "name": "authentication" }
Fallback
If no topics found from title keywords:
- Try keywords from item description
- Try removing suffixes (-patterns, -conventions)
- Continue to file exploration without Kodex context
Display found topics before proceeding to "Check project state".
Process
-
Check project state:
- Read relevant files (code, docs, configs)
- Check recent commits for context
- Understand existing architecture
-
Gather requirements context:
- Review any existing specs or requirements
- Understand stakeholder expectations
- Identify constraints and dependencies
-
Form initial item list:
- Create a preliminary list of topics to discuss
- Group related items together
- Prioritize by importance/dependencies
Single-Item Mode (EXPLORING)
When
currentItem is set in collab-state.json:
- Read relevant files based on item description
- Check git history for related changes
- Gather context specific to this item only
- Do NOT expand scope beyond this item
Exit Criteria
- Context gathered from relevant files
- Initial list of items/topics formed
- Ready to discuss items one at a time
Checkpoint: Current State Diagram
REQUIRED before proceeding to CLARIFYING:
Create a diagram showing the current state relevant to this work:
Tool: mcp__plugin_mermaid-collab_mermaid__create_diagram Args: { "project": "<cwd>", "session": "<session>", "name": "current-state-item-N", "content": <flowchart showing existing components/flow> }
Display: "Current state diagram: [previewUrl]"
If the work item doesn't involve architecture/flow (e.g., pure text changes), create a simple diagram showing the file(s) being modified:
graph LR A[File: path/to/file.md] --> B[Section being modified]
Transition to CLARIFYING
Prerequisites:
- Read relevant files/context
- Formed initial list of items
Announce: "I've gathered context. Now let me discuss each item with you one at a time."
Completion
At the end of this skill's work, call complete_skill:
Tool: mcp__plugin_mermaid-collab_mermaid__complete_skill Args: { "project": "<cwd>", "session": "<session>", "skill": "brainstorming-exploring" }
Handle response:
- If
: Invoke skill: collab-clearaction == "clear" - If
is not null: Invoke that skillnext_skill - If
is null: Workflow completenext_skill