Skip to content

claude-code.md

Latest commit

 

History

History
75 lines (55 loc) · 1.68 KB

claude-code.md

File metadata and controls

75 lines (55 loc) · 1.68 KB

Synapto + Claude Code

Setup

  1. Install Synapto and initialize the database:
pip install synapto
createdb synapto && psql -d synapto -c "CREATE EXTENSION vector;"
synapto init
  1. Add to your Claude Code MCP config.

Recommended (auto-updates via uvx) — add to ~/.claude/.mcp.json:

{
  "mcpServers": {
    "synapto": {
      "command": "uvx",
      "args": ["synapto", "serve"]
    }
  }
}

Per-project with tenant isolation (.claude/settings.json in your repo root):

{
  "mcpServers": {
    "synapto": {
      "command": "uvx",
      "args": ["synapto", "serve"],
      "env": {
        "SYNAPTO_DEFAULT_TENANT": "my-project"
      }
    }
  }
}

Why uvx? It resolves the latest version from PyPI each time Claude Code starts the server. No manual pip install --upgrade needed.

  1. Restart Claude Code. Synapto tools will appear in your tool list.

Usage Examples

Claude Code will automatically use Synapto tools. You can also ask directly:

  • "Remember that Hermes uses the outbox relay pattern for Kafka"
  • "What do you know about our Kafka architecture?"
  • "What depends on the agent.trigger topic?"
  • "Show me memory stats"

Migrating from MEMORY.md

synapto import ~/.claude/projects/your-project/memory/MEMORY.md --format markdown -t your-project

This imports each ## Section as a separate stable memory with full semantic search.

Multi-Project Setup

Use per-project configs with different tenants to keep memories isolated:

# In project A's .claude/settings.json
"env": { "SYNAPTO_DEFAULT_TENANT": "project-a" }

# In project B's .claude/settings.json
"env": { "SYNAPTO_DEFAULT_TENANT": "project-b" }