contextware

open-memory-local-manager

0
0
# Install this skill:
npx skills add contextware/skills --skill "open-memory-local-manager"

Install specific skill from multi-skill repository

# Description

>-

# SKILL.md


name: open-memory-local-manager
description: >-
Advanced skill for managing memories in OpenMemory Local. Provides automated
connectivity verification, search, and storage patterns.
version: 2.0.0
mcp-servers:
- open-memory-local
allowed-tools:
- open-memory-local_openmemory_store
- open-memory-local_openmemory_list
- open-memory-local_openmemory_query
tags: [memory, local, mcp, search]


OpenMemory Local Manager

This skill provides a robust workflow for interacting with an open-memory_local MCP server. It includes patterns for connectivity health checks, semantic search, and structured memory storage.

MCP Server Requirements

This skill requires the open-memory-local MCP server.

Connection Details:
- URL: Typically http://localhost:8082/mcp
- Transport: HTTP with server-sent events (SSE) support.

Authentication (Static Header)

[!IMPORTANT]
This server uses Static Header Authentication (API Key). It does NOT use OAuth2.

Instruction for Agent:
1. Ignore OAuth: Even if the platform suggests OAuth or returns requires_oauth_config, do not ask the user for a Client ID, Client Secret, or Redirect URI.
2. Request API Key: Always ask the user for their x-api-key.
3. Store Credentials: Use the store_api_key_credentials tool with headers: { "x-api-key": "[KEY]" }.
4. Connect: Call connect_to_mcp_server only after storing the key.

Required Header:

x-api-key: [YOUR_API_KEY]

Header Compatibility:
When calling this server directly, clients must include the following headers for compatibility with the SSE transport:

Accept: application/json, text/event-stream
x-api-key: [YOUR_API_KEY]

Workflow

Phase 1: Connectivity & Auth Verification

Before performing operations, verify the server is accessible and the API key is valid.

  1. Check Connection: Use the bundled scripts/check-connection.cjs script.
    • Usage: node scripts/check-connection.cjs <URL> --api-key=<KEY>
    • Note: If you don't have an API key yet, ask the user for it first. Do not attempt OAuth.
  2. Handle Response:
    • status: success: Proceed to Phase 2.
    • error: authentication_required: Ask the user for the correct x-api-key. Do not offer OAuth as an alternative.
    • status: error: Verify the local service is running on the specified port. If using a gateway URL (e.g., mcpgateway.online), ensure the URL is correct.

Phase 2: Memory Operations

Searching Memories

Use openmemory_query for semantic search.
- When to use: When the user asks "What do I know about X?" or "Find memories related to Y".
- Parameters: Use a descriptive query string. Setting k (limit) is recommended for large datasets.

Listing Recent Memories

Use openmemory_list for a broad overview.
- When to use: When the user asks "Show my recent memories" or "What is in my memory?".

Storing Memories

Use openmemory_store to persist new facts.
- Best Practice: Include tags (array of strings) to help with later categorization.
- Metadata: Use the metadata object to store source information (e.g., content_type: "conversation") if available.

Platform Adaptation Notes

[!NOTE]
Different agent platforms handle the SSE (Server-Sent Events) transport differently.

For Agents with Direct HTTP Access

Always ensure the Accept header includes text/event-stream. The server may fail if only application/json is requested.

For Agents with Global MCP Config

Ensure the x-api-key is injected into the server's environment or configuration.

Bundled Scripts

scripts/check-connection.cjs

A portable Node.js script to verify the MCP server health.

Input:
- URL (positional)
- --api-key (optional flag)

Output: JSON with status and serverInfo.


Troubleshooting

"API key required" / Authentication Error
→ The server requires the x-api-key header. Provision this from the environment or ask the user.

"Not Acceptable" / Header Error
→ The server requires the Accept: text/event-stream header. Ensure your transport layer is configured correctly.

"Connection timed out"
→ The local server at localhost:8082 (or your configured port) is likely not running or blocked by a firewall.

# Supported AI Coding Agents

This skill is compatible with the SKILL.md standard and works with all major AI coding agents:

Learn more about the SKILL.md standard and how to use these skills with your preferred AI coding agent.