brain-explore

Installation
SKILL.md

Brain Explore — Interactive Graph Traversal

ALL responses MUST be in Thai language.

Purpose

เดินตาม graph จาก node หนึ่งไปอีก node ผ่าน relationships — ค้นหาแบบ "ตามเส้น" ไม่ใช่แบบ keyword

Steps

  1. Find starting node

    • Call mcp__graph-brain__search-knowledge query="{argument}" limit=5
    • If multiple matches → show numbered list, let user pick
    • If exact match → auto-select
  2. Load starting node

    • Call mcp__graph-brain__get-knowledge noteId="{id}"
    • Display brief summary of node content (max 5 lines)
  3. Explore connections

    • Call mcp__graph-brain__explore-graph nodeId="{id}" depth=1
    • Parse response to extract connected nodes and relationship types
  4. Display interactive map (Thai)

    🧭 Exploring: [{Note Title}]
    📍 Path: {breadcrumb if not first hop}
    ─────────────────────────────────────
    {Brief content summary — max 5 lines}
    ─────────────────────────────────────
    
    🔗 Connections ({N} total):
    
    📄 Notes (LINKS_TO):
      [1] → {Connected Note Title 1}
      [2] → {Connected Note Title 2}
    
    🏷️ Tags (TAGGED):
      [3] → #{tag-1}
      [4] → #{tag-2}
    
    📁 Folder:
      [5] → /{folder-path}/
    
    🏗️ Project:
      [6] → {Project Name}
    
    ─────────────────────────────────────
    เลือก:
    [เลข] → explore node นั้นต่อ
    [B]   ← กลับ node ก่อนหน้า
    [S]   🔍 search ใหม่
    [Q]   หยุด explore
    
  5. Handle navigation

    • Number selected → load that node (Step 2), explore connections (Step 3), display (Step 4)
    • [B] Back → return to previous node in breadcrumb stack
    • [S] Search → ask for new keyword, go to Step 1
    • [Q] Quit → end exploration, show summary
  6. Maintain breadcrumb

    • Track path: Node A → Node B → Node C
    • Display breadcrumb at top of each hop
    • Max depth: 10 hops (warn user if approaching limit)
  7. End exploration summary

    🧭 Exploration Summary
    ═══════════════════════════════════════
    📍 Path taken: {A} → {B} → {C} → {D}
    📊 Nodes visited: {N}
    🔗 Unique relationships seen: {M}
    
    💡 ลองต่อ:
      /brain-explain {most interesting node} — อธิบายแบบละเอียด
      /brain-history {node with versions} — ดูประวัติเปลี่ยนแปลง
    
  8. Write activity log

    • Append entry to .brain/activity-log.json
    {
      "timestamp": "<ISO 8601 UTC>",
      "session_id": "<$CLAUDE_SESSION_ID or date-based>",
      "command": "brain-explore",
      "args": "<starting keyword>",
      "project": "<project-name from cwd>",
      "status": "completed",
      "details": {
        "start_node": "<starting note title>",
        "nodes_visited": N,
        "path": ["<node titles in order>"],
        "hops": N
      }
    }
    

Options (via argument flags)

Flag Effect
--depth N Show connections up to N hops at once (default: 1)
--links Filter: show only LINKS_TO connections
--tags Filter: show only TAGGED connections
--project {name} Filter: show only nodes in specified project
Related skills

More from mounchons/agentmarketplace

Installs
1
First Seen
Apr 7, 2026