agentmemory-agents skill for Claude Code: wire agentmemory into your coding agent

Teaches Claude to connect agentmemory, a persistent memory server, to your coding agent with the agentmemory connect command.

Installs: 11.3K; on skills.sh
Stars: 30.1K; on GitHub
Files: 2; besides SKILL.md
Updated: Oct 2026; last repo commit

agentmemory-agents is a reference skill that ships with agentmemory, a memory server for coding agents. Its description says Claude should use it when installing agentmemory into a specific agent, when asked which agents are supported, or when a connect adapter writes the wrong config path. It is not user-invocable, so Claude loads it by itself and there is no slash command.

The core is agentmemory connect <agent>, which merges the memory server into the agent's config and keeps your existing servers. REFERENCE.md lists 22 supported agents and TROUBLESHOOTING.md covers checks when tools do not show up. The skill needs the agentmemory CLI and server, which the install box explains.

Good for

Developers setting up agentmemory in Claude Code, Cursor, Codex or Gemini CLI
Teams running several coding agents that should share one memory
Anyone fixing a connect adapter that wrote the wrong config path

Install agentmemory-agents

Claude Code

Copy this prompt.

Install the agentmemory-agents skill in Claude Code. Run this command:

/plugin marketplace add rohitg00/agentmemory
/plugin install agentmemory

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Requires Node.js 20 or newer.

Start the memory server first, in its own terminal: npx -y @agentmemory/agentmemory@latest. agentmemory-agents is a reference for connecting agentmemory to other agents, so it does little without the server.

From the project README. Source.

What's in the skill

When it loads

How agentmemory wires into host coding agents via the connect command. Use when installing agentmemory into a specific agent, when asked which agents are supported, or when a connect adapter writes the wrong config path.

Sections in SKILL.md

Quick start
Workflow
Notes
See also
Reference

Key facts

Author: rohitg00 (agentmemory)
Repository: rohitg00/agentmemory
License: Apache-2.0 (crawl data)
Plugin: agentmemory (17 skills in plugin/skills)
Invoked by: Claude only (user-invocable: false)
Installs on skills.sh: 11,289 (Nov 2, 2026)
GitHub stars (repo): 30,069 (Nov 2, 2026)
Last repo commit: Oct 29, 2026
Files: 3 files including SKILL.md (REFERENCE.md, TROUBLESHOOTING.md)

What it includes

Merges the memory server into a host agent's config and keeps existing servers.
Run it as agentmemory connect claude-code (or cursor, codex, gemini-cli). Restart the host or reload its MCP list afterwards.
Runs code · agentmemory connect <agent>
Restart the host, list its tools, then do an authorized save and recall check.
Read · Verify the result
One devin plugins install command adds the skills, MCP server and capture hooks.
Runs code · Devin plugin install
connect pi and omp use --with-mcp, and connect dsh uses --with-hooks.
Runs code · Pi, Oh My Pi, DeepSeek Harness
The connector configures MCP. Lifecycle capture comes from the Cursor plugin in Settings.
Cursor setup
A skills CLI command with --agent hermes-agent or openclaw installs their skills.
Runs code · Skills for Hermes and OpenClaw
22 adapters with their config files and protocol notes.
REFERENCE.md adapter table
Checks agentmemory status, the tools setting and the AGENTMEMORY_URL value.
Read · TROUBLESHOOTING.md

From the official docs

Permissions

Medium risk

The skill holds instructions, but the connect command it runs edits your agent's config, and the plugin's hooks record prompts and tool use to a local memory server. Back up your config and check what each hook captures.

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

connect edits your agent's config files
agentmemory connect merges into the host's config, such as ~/.gemini/config/mcp_config.json or ~/.factory/mcp.json. It keeps existing servers, but back up the file first. github.com
Hooks record your prompts and tool use
The plugin hooks run node scripts on events like UserPromptSubmit and PostToolUse. The README says prompts are privacy-filtered and stored by the local server. github.com
It starts a local server
Setup runs npx -y @agentmemory/agentmemory@latest. The README says the API and viewer bind to loopback by default. github.com
Local embeddings download a model
With EMBEDDING_PROVIDER=local, the README says the first embedding request downloads the Xenova/all-MiniLM-L6-v2 model. github.com

Technical details

Talks to	localhost:3111 (agentmemory server, loopback by default). localhost:3113 (live viewer)

Questions

How do I install agentmemory-agents?

Use the install box above. The skill is only useful with the agentmemory server running, so start it first as the box notes. The Claude Code plugin route also adds 12 hooks, 17 skills and an MCP bridge.

What is agentmemory?

The README describes it as persistent memory across coding sessions and agents, for Claude Code, Copilot CLI, Cursor, Gemini CLI, Codex CLI and others.

How do I use this skill?

You don't call it. Ask Claude to connect agentmemory to your agent, or ask which agents are supported, and Claude loads it. The SKILL.md workflow detects the calling agent, runs the connect command and checks the tools list.

Which agents does it cover?

REFERENCE.md lists 22 adapters, including Claude Code, Cursor, Codex CLI, GitHub Copilot CLI, Gemini CLI, OpenCode, Cline, Continue, Zed, Warp, Kiro, Hermes Agent and Pi.

Does it need an API key?

SKILL.md names no API key. The README describes a keyless mode that uses BM25 search, and optional LLM or embedding providers you can add later.

Is it safe?

The skill folder has three Markdown files and no scripts. The connect command it runs edits your agent's config files, and the plugin adds hooks that record your prompts and tool use to a local server. The permissions section lists what each part does.

Does it work on Windows?

SKILL.md says native Windows supports connect for copilot-cli. Other automated connectors need macOS, Linux or a host in the same WSL2 environment.

Information

updated Nov 2, 2026

Repository: rohitg00/agentmemory
Folder: plugin/skills/agentmemory-agents

rohitg00/agentmemory skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 7

agentmemory README · github.com
SKILL.md · github.com
Marketplace manifest · github.com
REFERENCE.md · github.com
TROUBLESHOOTING.md · github.com
skills.sh listing · skills.sh
Repository · github.com