wasurenagusa

Provides persistent memory for Claude Code, automatic rule injection, and autonomous spec updates to accelerate productive conversations.

typescript

GitHub Stars

typescript

Language

4 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "tsutushi0628-wasurenagusa-mcp": {
      "command": "node",
      "args": [
        "/path/to/wasurenagusa-mcp/dist/index.js"
      ],
      "env": {
        "MEMORY_DIR": ".wasurenagusa",
        "GEMINI_API_KEY": "YOUR_GEMINI_API_KEY",
        "SLACK_WEBHOOK_URL": "YOUR_SLACK_WEBHOOK_URL"
      }
    }
  }
}

wasurenagusa MCP Server provides an autonomous memory and automation layer for Claude Code conversations. It automatically injects memory, rules, and past decisions at the start of each session, updates documentation and specifications during off-hours, and autonomously runs tasks to optimize token usage and keep knowledge fresh. You interact with Claude Code as usual, while the MCP handles behind-the-scenes memory management and self-improving workflows.

How to use

You install and run the MCP server, connect it to Claude Code via the provided CLI integration, and then you can rely on it to remember your project rules, past decisions, API endpoints, and constraints across sessions. The system automatically injects memory at session start, analyzes conversations after each session, and updates the spec and related documentation as needed. You do not need to perform manual memory operations.

How to install

Prerequisites: Node.js 18 or newer, Claude Code CLI, and a Gemini API key.

git clone git@github.com:tsutushi0628/wasurenagusa-mcp.git
cd wasurenagusa-mcp
npm install
npm run build

npm link

Additional setup steps

Prepare environment variables and memory storage location.

cp .env.example .env

GEMINI_API_KEY=your_gemini_api_key_here
MEMORY_DIR=.wasurenagusa
SLACK_WEBHOOK_URL=                    # Optional: enable Slack notifications

Register MCP server with Claude Code

claude mcp add wasurenagusa -- node /path/to/wasurenagusa-mcp/dist/index.js

Hook configuration and memory ingestion

Configure memory hooks so the server automatically injects memory at session start and analyzes conversations at stop.

{
  "hooks": {
    "SessionStart": [{"hooks": [{"type": "command","command": "wasurenagusa-context","timeout": 5}]}],
    "Stop": [{"hooks": [{"type": "command","command": "wasurenagusa-analyze","timeout": 30}]}]
  }
}

Validation steps

Make sure the MCP server is properly registered and the memory paths are writable. The first session will create a memory directory and generate an owner profile template that you can customize to influence autonomous task execution.

Security and maintenance

Store memory data in the configured memory directory and keep it ignored by your version control system to avoid leaking project-specific memory. Regularly review the owner profile and memory rules to ensure they align with your security posture.

Available tools

memory_get_context

Fetches the current config and dont rules for the session to provide context to the AI autonomously.

memory_search

Lightweight index search that returns only IDs, titles, and tags for quick discovery.

memory_get_detail

Retrieves full details for a specific memory entry by its ID.

memory_save

Explicitly save a memory entry manually if needed.

memory_delete

Delete memory entries by ID, with support for batch deletion.

task_submit

Submit autonomous tasks that the scheduler will execute using Claude CLI as a subprocess.

task_status

Query the current status of autonomous tasks.

task_action_list

List available actions that autonomous tasks can perform.

project_init

Manually initialize project settings, including quality standards and decision criteria.

Built by

VeilStrat

AI signals for GTM teams