@aradenta.labs/centmem-skills

Frictionless installer for cent-mem skills & slash commands across AI agent harnesses.

centmem-skills is a zero-dependency, cross-platform installer that equips your AI coding agents with cent-mem shared memory and intuitive /centmem slash command capabilities.

Prerequisites: Install centmem CLI

Before connecting your AI agents, ensure the centmem Go binary is installed and initialized on your machine.

1. Install centmem

Option A: Pre-built Binary (macOS & Linux) Download the latest release for your OS and architecture from GitHub Releases:

# Example for macOS (Apple Silicon):
curl -fsSL -o centmem https://github.com/aradenta-labs/cent-mem/releases/latest/download/centmem-darwin-arm64
chmod +x centmem
sudo mv centmem /usr/local/bin/

(Pre-built binaries available for darwin-arm64, darwin-amd64, linux-amd64, and linux-arm64.)

Option B: Install with Go (Go 1.22+)

go install -tags fts5 github.com/aradenta-labs/cent-mem/cmd/centmem@latest

2. Initialize centmem (Run once per machine)

centmem init

This creates the local database (~/.centmem/) and downloads the local ONNX embedding model (~100 MB).

Quickstart: Install Agent Skills

Run the installer directly from your project root:

npx @aradenta.labs/centmem-skills

What It Does

When you run npx @aradenta.labs/centmem-skills:

1. Installs Detailed Skill Artifacts: Auto-discovers and installs SKILL.md, comprehensive references, workflow examples, and helper scripts into standard AI skill locations:

   • Project-Level Agents Skills (./.agents/skills/centmem)
   • User-Level Global Agents Skills (~/.agents/skills/centmem)
   • Claude Code (./.claude/skills/centmem, ~/.claude/skills/centmem)
   • Cursor (~/.cursor/rules/centmem)
   • Google Antigravity (~/.gemini/antigravity/skills/centmem)
   • OpenAI Codex (~/.codex/skills/centmem)
   • Trae (~/.trae/skills/centmem)
   • Hermes Agent (~/.hermes/skills/centmem)
   • DeepSeek Harness (~/.deepseek/skills/centmem)
   • Global Config (~/.config/skills/centmem)

2. Automated Memory Loop Injection in AGENTS.md: Guarantees that AGENTS.md is updated (or created) in your project root, and injects into any existing rule files (CLAUDE.md, .cursorrules, .github/copilot-instructions.md) to establish the standard memory loop:

   • READ: Run centmem recall "<task context>" to load prior context.
   • DO: Execute the task.
   • UPDATE: Run centmem put or centmem set to persist decisions and learnings.

3. Smart /centmem Slash Command Support: Registers routing instructions that allow you to interact with memory using simple prompts like /centmem review recent architectural decisions (triggers recall), /centmem remember we use PostgreSQL for this service (triggers put/set), or /centmem configure llm backend to ollama (triggers config get/set).

CLI Usage & Options

# Default installation
npx @aradenta.labs/centmem-skills

# Install with explicit repository source
npx @aradenta.labs/centmem-skills add https://github.com/aradenta-labs/cent-mem --skill centmem

# Preview targets and planned actions without touching disk
npx @aradenta.labs/centmem-skills --dry-run

# List detected instruction files and target directories
npx @aradenta.labs/centmem-skills --list

# Cleanly remove installed skill directories and injected instructions
npx @aradenta.labs/centmem-skills --uninstall

Options

| Option | Description | |---|---| | --global | Install globally to user-level harness directories (~/.claude/, ~/.cursor/, etc.) | | --project | Install locally to project-level harness directories (./.agents/, ./.claude/, etc.) | | --yes, -y | Skip interactive prompts and accept all detected defaults | | --skill <name> | Custom skill name (default: centmem) | | --dry-run | Simulate actions without modifying the filesystem | | --list | Display target install paths and detected instruction files | | --uninstall | Cleanly remove skill files and instruction blocks | | --cwd <path> | Override target working directory | | --home <path> | Override target home directory | | -h, --help | Show command help |

About cent-mem

cent-mem is a fast, local-first shared memory store for AI agents powered by SQLite, vector search, BM25 full-text search, and a built-in cognitive memory agent engine.

• GitHub Repository: github.com/aradenta-labs/cent-mem
• Documentation: cent-mem Docs
• License: MIT