MCP Tools Directory
← Browse integrations

Developer tools

Sugar

Free account · sign-in required

Contact Sugar

Loading secure sign-in…

Tools for autonomous AI agents

About this integration

Provides local, persistent project and global memory that coding agents can search and update through MCP.

Transport
stdio
Authentication
none
Initial setup
none
Runtime
unattended
Evidence
documented
Package
sugarai
Last compatibility test
Not independently tested

Connect your agent

sugarai

Publisher documentation and source manifest reviewed; the package, local databases, optional models, agent integrations, and autonomous task features were not executed or independently security-audited. This review counts the sugarai executable package once even though it exposes memory and optional task MCP modes. Local memory may contain sensitive project context and should be protected accordingly.

Capabilities: Store and search project-local and global coding-agent memories, Recall project context, preferences, guidelines, and recent memories, Optionally expose a separate local task-queue MCP mode

Connected profiles

Additional details

Source artifact

sugarai

Source ↗ · Checked 2026-09-19
Source manifest version

3.10.2

Source ↗ · Checked 2026-09-19
Unattended configuration

A coding agent can access Sugar's local project and global memory automatically through the sugar mcp memory server; an optional task server is a mode of the same executable artifact.

Source ↗ · Checked 2026-09-19
Authentication and prerequisites

The local-first memory feature is documented as fully offline-capable with no API keys; optional agent-driven task execution can have separate provider requirements outside the MCP memory server.

Source ↗ · Checked 2026-09-19
Connection configuration

The documented Claude Code connection launches sugar mcp memory as the local MCP command; sugar mcp tasks is an optional second mode of the same package.

Source ↗ · Checked 2026-09-19
Review scope

Publisher documentation reviewed; package and integration endpoint not executed or independently security-audited.

Source ↗ · Checked 2026-09-19