Quick Start

1. Initialize a Knowledge Base

export PIEKBS_KB=/path/to/your-kb

piekbs init    # scaffold KB dirs and copy schema/templates
piekbs serve   # start server: MCP + Web UI + file watcher

On macOS, double-click PieKBS.app to launch as a menubar icon.

2. Configure MCP in Your Agent

HTTP mode (recommended — one process shared by all agents):

Add to ~/.claude.json under mcpServers:

{
  "mcpServers": {
    "piekbs": {
      "type": "http",
      "url": "http://127.0.0.1:8766/mcp",
      "headers": {
        "x-api-key": "${PIEKBS_API_KEY}"
      }
    }
  }
}

x-api-key corresponds to server.api_key in config.yaml. Omit headers if no api_key is set.

stdio mode (for hosted environments):

{
  "mcpServers": {
    "piekbs": {
      "type": "stdio",
      "command": "/path/to/piekbs",
      "args": ["stdio"],
      "env": {
        "PIEKBS_KB": "/path/to/your-kb"
      }
    }
  }
}

3. Add Content

Drop any Markdown file into raw/. The watcher automatically distills it.

cp my-notes.md $PIEKBS_KB/raw/
# watcher detects the file and triggers distill + reindex automatically

4. Search

piekbs search "your query"

Or let your agent use the MCP tools:

kb_search("keyword A")          → discover relevant documents
kb_search("keyword B")          → cover a different angle
kb_page(["id1", "id2"])         → deep-read the most relevant ones