brain-base

Query and persist local documents through a structured knowledge base workflow.

1|Updated Apr 12, 2026
One-click install
npx skills add https://github.com/mengsi16/brain-base --skill brain-base
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: brain-base
Source: https://github.com/mengsi16/brain-base/tree/main/brain-base-skill
Command: npx skills add https://github.com/mengsi16/brain-base --skill brain-base

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Brain-base enables Claude Code / Codex agents to route questions to a structured local knowledge-base workflow and decide when to ingest new documents, enabling more accurate, auditable Q&A over time.

Core Features & Use Cases

  • Two-entry-point design: qa-agent for local question answering and get-info-agent for on-demand web research, and upload-agent for local document ingestion.
  • Controlled prompts & routing: clear guidelines on how to craft prompts, when to fetch, and how to solidify feedback into crystallized data.
  • Lifecycle for knowledge: from local docs to raw/Chunk storage to Milvus indexing and crystallized planning.

Quick Start

Invoke brain-base by calling claude -p with a question for the qa-agent or a file path for the upload-agent, using --dangerously-skip-permissions to automate the workflow.

Frequently Asked Questions about brain-base

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I use Claude Code to query a local knowledge base?

To query a local knowledge base with Claude Code, you invoke the qa-agent via the claude -p command. This routes your question to a structured workflow that retrieves answers from indexed local documents.

Do I need Milvus to run local document ingestion with Claude Code?

Yes, Milvus is required for local document ingestion and retrieval. It serves as the vector database backend for indexing chunks and enabling structured question answering over your persisted data.

What is the best way to ingest local documents into a knowledge base for AI agents?

The best way to ingest local documents for AI agents is using the upload-agent entry point. You pass the file path to the agent, which automates routing, chunking, and indexing into Milvus for later retrieval.

Why does the local knowledge base workflow require the dangerously-skip-permissions flag?

The dangerously-skip-permissions flag is required to automate the knowledge base workflow without manual confirmation prompts. It allows Claude Code to autonomously route questions and execute document ingestion commands.

Can I fetch web research on demand while querying a local document knowledge base?

Yes, you can fetch web research on demand using the get-info-agent. This entry point triggers automatically when the local question answering workflow needs external information beyond the ingested documents.