nexus-memory AGENTS.md

Installation instructions for Nexus Memory, a shared memory layer for AI agents such as Hermes, OpenClaw, Claude Code, and Cursor. They explain how to install its Hermes plugin and verify the configuration.

In plain words
What is it for?
Use them to install the Nexus memory plugin for Hermes, activate it in the configuration, verify the selected provider, and restart the Hermes gateway.
Why use it?
They give an agent or developer a documented way to connect Hermes to shared memory instead of setting up each installation step from scratch.

Instructions file for CodexOpenCode

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/neboy72/nexus-memory/agents-md
Clone the repo
git clone --depth 1 https://github.com/Neboy72/nexus-memory

Made for: Codex, OpenCode.

Per session 5,824 This file is loaded in full into every session.
When invoked 5,824 The same file — it is already loaded in full.
Security scan C 1 finding. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.05824 $0.05824
Opus 5 $0.02912 $0.02912
Sonnet 5 $0.01165 $0.01165
Haiku 4.5 $0.00582 $0.00582

Measured 2d ago against content hash cb2a465326ac, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade C, and why

nexus-memory AGENTS.md scanned grade C with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Recursive force deletehighDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

Before any destructive operation (`rm -rf`, `drop`, `kill -9`, `recreate_collection`), the guardrail checks Qdrant for stored protection rules (category=`rule` with protection keywords like "never delete", "protected", "
AGENTS.md · 563 lines

How it starts

The opening of the file, as written. The whole thing — 563 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Nexus Memory — AGENTS.md (AI Installation Guide)

Tell your AI agent: "Read https://raw.githubusercontent.com/Neboy72/nexus-memory/main/AGENTS.md and follow the installation instructions."

Overview

Nexus Memory is a universal memory layer for AI agents. One memory for all your agents — Hermes, OpenClaw, Claude Code, Cursor, or any MCP-compatible agent.

If you use Hermes Agent, install Nexus Memory as a native MemoryProvider plugin. This gives you direct Qdrant access with zero MCP overhead — Hermes reads/writes memories as part of its core loop.

One-Command Install

cd ~/nexus-memory
./scripts/install_hermes_plugin.sh

This script:

  1. Symlinks the nexus plugin into ~/.hermes/hermes-agent/plugins/memory/
  2. Sets memory.provider to nexus in Hermes config
  3. Verifies the installation

Manual Setup

Or, run these commands yourself:

# Link the plugin
ln -s ~/nexus-memory/plugins/memory/nexus ~/.hermes/hermes-agent/plugins/memory/nexus

# Activate it
hermes config set memory.provider nexus

Verify

hermes config get memory.provider
# → nexus

Restart Hermes Gateway. Nexus tools appear as nexus_recall, nexus_remember, nexus_forget.

If you use OpenClaw, install Nexus Memory as a native memory plugin. This gives you Auto-Recall (memories injected before every turn) and Auto-Capture (facts extracted after every turn) — all powered by your local Qdrant.

One-Command Install

cd ~/nexus-memory
./scripts/install_openclaw_plugin.sh

This script:

  1. Detects your OpenClaw installation
  2. Adds the plugin to plugins.load.paths in ~/.openclaw/openclaw.json
  3. Sets plugins.slots.memory = "nexus-memory"
  4. Auto-detects your embedding provider (Voyage > OpenAI > Google > Jina > Ollama)
  5. Restarts OpenClaw gateway

Manual Setup

Or, run these commands yourself:

  1. Add plugin path to ~/.openclaw/openclaw.json:
{
  "plugins": {
    "load": {
      "paths": ["/path/to/nexus-memory/plugins/openclaw"]
    },
    "slots": { "memory": "nexus-memory" },
    "entries": {
      "nexus-memory": {
        "enabled": true,
        "hooks": {
          "allowPromptInjection": true,
          "allowConversationAccess": true
        },
        "config": {
          "qdrantUrl": "http://localhost:6333",
          "collection": "nexus",
          "embedding": {
            "provider": "voyage",
            "model": "voyage-4",
            "apiKey": "${VOYAGE_API_KEY}"
          },
          "autoRecall": true,
          "autoCapture": true,
          "maxRecallResults": 10,
          "accessLevel": "private"
        }
      }
    }
  }
}

Read the full file on GitHub · 563 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 563 lines · 5,824 tokens per session scan C cb2a465326ac

Subscribe to this mod's changes

nexus-memory AGENTS.md is an instructions file published in the GitHub repository Neboy72/nexus-memory (2 stars, last pushed 2d ago), licensed MIT. It adds 5,824 tokens to every session, about $0.0291 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.