hmem — Setup

What it does

This skill provides an interactive installer (hmem init) and manual setup instructions to install and configure hmem — a persistent-memory MCP server that integrates with agent tools (Claude Code, Gemini CLI, Cursor, OpenCode). It automates dependency installation, writes .mcp.json entries, creates a memory directory and default hmem.config.json, and deploys helper skill files and Claude hooks so agents can read and write long-term memory.

When to use it

Use this skill when you want to add persistent, structured memory to your AI agent workflow. It's appropriate during initial environment setup for a developer machine or CI environment, when migrating memory across machines, or when you need consistent hook-based checkpointing and exchange logging for agent sessions.

What's included

• Scripts: none in this SKILL.md package (has_scripts=false).
• References: none bundled.
• Instructions: step-by-step commands for hmem init, a non-interactive flag for CI, manual fallback steps (clone, build, configure .mcp.json), and a configuration reference for tuning memory behaviour and bulk-read selection. The SKILL.md also documents Claude-specific hook scripts and their behavior (UserPromptSubmit, Stop, SessionStart) and troubleshooting tips.

Compatible agents

Inferred compatible tools: Claude Code, Gemini CLI, Cursor, OpenCode, Windsurf. The skill explicitly documents Claude Code hooks and mentions other CLI tools in its compatibility list.