brain
Brain First — Primary Knowledge Query
ALL responses MUST be in Thai language regardless of input language.
Execution Flow
Step 1: Search Brain (fast path)
- Call
mcp__graph-brain__search-knowledgequery="{user's question}" limit=10 - Call
mcp__graph-brain__search-by-tagswith extracted keywords as tags - For top 3-5 results, call
mcp__graph-brain__get-knowledgeto load full content - Follow
[[wiki links]]in loaded notes — load linked notes for complete context (max 3 hops) - If results < 3 and at least 1 result exists → use
mcp__graph-brain__explore-graphnodeId="{best-result-id}" depth=2 to find connected knowledge through relationships
Step 2: Evaluate Completeness
Rate the brain results against the user's question:
- Complete (brain answers the question fully) → go to Step 4
- Partial (some info found but gaps exist) → go to Step 3
- Empty (nothing found) → go to Step 3
Step 3: Read Codebase (slow path — only if needed)
Display to user:
🧠 Brain: พบข้อมูล {N} ชิ้น {complete|บางส่วน|ไม่พบ}
📂 กำลังอ่านเพิ่มจาก codebase...
- Use Explore agent or direct file reads to find missing information
- Focus only on the GAP — do not re-read what brain already provided
- After reading, combine brain knowledge + codebase findings
Step 4: Respond in Thai
Present the answer with:
- Clear structure (headers, tables, flow diagrams as appropriate)
- Source labels:
[Brain]for brain-sourced info,[Code]for codebase-sourced info - Relevant file paths if referencing specific code
- Relationship context: if explore-graph was used, show key connections found (e.g., "เชื่อมกับ: [[Note A]], [[Note B]]")
Step 5: Offer to Save (only if Step 3 was executed)
If new information was found from codebase:
💡 พบข้อมูลใหม่จาก codebase ที่ยังไม่มีใน Brain
ต้องการบันทึกไหม? พิมพ์ /brain-save เพื่อเก็บ
Connection Failure
If Graph Brain is unreachable:
⚠️ Graph Brain เชื่อมต่อไม่ได้ — อ่านจาก codebase โดยตรง
Then read from codebase and answer normally. Do NOT block.
Examples
/brain checker ทำอะไรบ้าง
/brain วิธีนำเข้าข้อมูลจาก Excel
/brain workflow ตั้งแต่รับงานถึงส่งประกัน
/brain database connection ใช้อะไรบ้าง
/brain Lucky job ต่างจากงานปกติอย่างไร
More from mounchons/agentmarketplace
ui-mockup
Create and edit UI Mockup/Wireframe from System Design Documents. Supports ASCII wireframes, component specifications, design tokens, and responsive design specs
13system-design-doc
Create standardized system design documents. Supports Reverse Engineering from codebase with Mermaid diagrams, ER Diagram, Flow Diagram, DFD, Sitemap, Sequence Diagram, and Data Dictionary
8dotnet-dev
Expert .NET Core development - ASP.NET Core MVC, Entity Framework Core, Clean Architecture, Repository/Unit of Work patterns, .NET Aspire รองรับ PostgreSQL และ SQL Server
7long-running-agent
Harness สำหรับ AI Agent ทำงานข้าม context windows รองรับ multi-session, feature tracking, progress logging, incremental development, epic grouping, subtask tracking, acceptance criteria และ integration กับ ui-mockup, system-design-doc, dotnet-dev skills
2long-running
Harness for AI Agents working across context windows — supports multi-session, feature tracking, progress logging, incremental development, epic grouping, subtask tracking, acceptance criteria, model assignment, review system, flows, state contracts, component requirements, CRITICAL RULES enforcement, Verification Pipeline, and integration with ui-mockup, system-design-doc, dotnet-dev skills
1brain-projects
View all projects in Brain — tech stacks, cross-project connections, and technology overview
1