mcp-memory-server 1.4.0

MCP server that exposes Project AI Memory (memory.md) as resources and tools for Cursor, Claude, and other MCP clients

mcp-name: io.github.NeetPatel/devops-mcp

DevOps MCP Server

MCP server that exposes Project AI Memory (memory.md) as resources and tools so AI clients (Cursor, Claude Desktop, etc.) can load DevOps/DevSecOps context.

Install

pip install mcp-memory-server

Run from a directory that contains memory.md, or set the path:

mcp-memory
# or
MCP_MEMORY_PATH=/path/to/memory.md mcp-memory

What it exposes

Memory (project policy):

• Resource memory://project-context – full contents of memory.md
• Tools – get_memory_section, get_security_checklist, get_code_standards

Execution tools (MCP tool names):

DevOps tool	MCP tool name	Implementation
Terraform	terraform.plan	CLI
Kubernetes	k8s.pods	CLI (kubectl)
Docker	docker.images	CLI
GitHub	github.action	GitHub API
Jenkins	jenkins.jobs	Jenkins API
Jenkins	jenkins.build	Jenkins API
Trivy	security.scan	CLI
Registry	registry.images	Registry V2 API
Drift	drift.check	Terraform CLI
K8s drift	k8s.drift	kubectl diff
DBA	dba.query	psql (CLI)

Install CLIs (kubectl, terraform, docker, trivy, psql, curl) via bash install-devops-tools.sh; see TOOLS.md. Env: GITHUB_TOKEN, JENKINS_URL (+ optional JENKINS_USER/JENKINS_TOKEN), REGISTRY_URL (+ optional REGISTRY_USER/REGISTRY_TOKEN) for private registries.

Load in Cursor (any system, like AWS Labs MCPs)

Add to Cursor MCP config (~/.cursor/mcp.json or Settings → MCP). Uses uvx so the server runs from PyPI without a local clone (same pattern as awslabs.cdk-mcp-server):

{
  "mcpServers": {
    "devops-mcp": {
      "command": "uvx",
      "args": ["mcp-memory-server@latest"],
      "env": {}
    }
  }
}

• Use mcp-memory-server for latest install, or mcp-memory-server@latest to pin to latest.
• Cursor uses your workspace as cwd, so put memory.md in the project root, or set a path in env:

"devops-mcp": {
  "command": "uvx",
  "args": ["mcp-memory-server@latest"],
  "env": {
    "MCP_MEMORY_PATH": "/path/to/your/memory.md"
  }
}

Requires uv (curl -LsSf https://astral.sh/uv/install.sh | sh).

Publish to PyPI (so others can use uvx)

From this repo, after tests pass:

uv run python -m build
uv run twine upload dist/*

Or with a GitHub workflow: build the package, then twine upload using a PyPI token. Once mcp-memory-server is on PyPI, anyone can use the Cursor config above on any machine.

Links

• Repository: https://github.com/NeetPatel/devops-mcp