Mnemon Memory

Persistent memory CLI for LLM agents. Store facts, recall past knowledge, link related memories, manage lifecycle.

Security Scan

The skill's description matches a CLI memory plugin, but its setup step installs third‑party code and injects persistent OpenClaw hooks/prompts that alter agent behavior — verify the upstream binary and exactly what files/hooks it writes before installing.

Purpose & Capability

Name/description (persistent memory CLI for LLM agents) aligns with the declared binary requirement and the install methods (Homebrew formula or go install of a mnemon repo). The SKILL.md's commands (remember, recall, link, etc.) are coherent for a memory tool.

Instruction Scope

The runtime instructions include a single 'mnemon setup --target openclaw' command that will create skill, hook, and plugin files under ~/.openclaw and prompt files under ~/.mnemon. This is expected for integration, but the setup injects an 'agent:bootstrap' behavioral guide and plugin hooks that will influence agent behavior on every message (remind/nudge). The SKILL.md does not show the exact files' contents; that makes it hard to audit what prompts/behavior are injected. The guide also explicitly tells the agent to use exec to run mnemon commands and warns not to store secrets.

Install Mechanism

Installers are a Homebrew tap (mnemon-dev/tap/mnemon) and a go install from github.com/mnemon-dev/mnemon@latest. Both are plausible for delivering a CLI but introduce supply-chain considerations: a third-party Homebrew tap is less vetted than an official formula, and 'go install @latest' pulls mutable code. No arbitrary URL downloads or extract steps are present.

Credentials

The skill does not request any environment variables or credentials. That is proportionate to its stated purpose. It will read/write files under the user's home (~/.openclaw, ~/.mnemon) which is reasonable for a local plugin and prompt store.

Persistence & Privilege

Although 'always' is false, the setup step explicitly modifies the OpenClaw installation by adding skills, hooks, and plugin config and instructs the user how to enable hooks that run on every message (remind/nudge). This grants persistent, high-impact influence over agent behavior. The skill will modify OpenClaw configuration (openclaw.json) and install agent:bootstrap hooks — actions with platform-wide consequences that should be reviewed before granting.

What to consider before installing

This skill appears to do what it says (a local CLI memory plugin) but it: (1) installs third‑party binaries (Homebrew tap or go install @latest), (2) runs a setup that writes files under ~/.openclaw and ~/.mnemon, and (3) injects bootstrap hooks/prompts that change agent behavior persistently. Before installing: (a) inspect the upstream project (GitHub repo and Homebrew tap) and prefer pinned releases or vetted packages, (b) run 'mnemon setup' in a throwaway/sandboxed environment or inspect the files it would write (ask the author for the exact file contents) so you can review injected prompts/hooks, (c) back up ~/.openclaw/openclaw.json and any agent configs so you can revert changes, and (d) avoid storing secrets in the memory store. If you cannot verify the source or review the setup output, treat installation as risky.

mnemon

Install & Configure

1. Install the binary

Homebrew (macOS / Linux):

brew install mnemon-dev/tap/mnemon

Go install:

go install github.com/mnemon-dev/mnemon@latest

2. Set up OpenClaw integration

mnemon setup --target openclaw --yes

This single command deploys all components:

• Skill → ~/.openclaw/skills/mnemon/SKILL.md
• Hook → ~/.openclaw/hooks/mnemon-prime/ (agent:bootstrap — injects behavioral guide)
• Plugin → ~/.openclaw/extensions/mnemon/ (remind, nudge, compact hooks)
• Prompts → ~/.mnemon/prompt/ (guide.md, skill.md)

Restart the OpenClaw gateway to activate.

3. Customize (optional)

Edit ~/.mnemon/prompt/guide.md to tune recall/remember behavior.

Plugin hooks are configured in ~/.openclaw/openclaw.json:

{
  "plugins": {
    "entries": {
      "mnemon": {
        "enabled": true,
        "config": {
          "remind": true,
          "nudge": true,
          "compact": false
        }
      }
    }
  }
}

Hook	Default	Description
remind	on	Recall relevant memories + remind agent on each message
nudge	on	Suggest remember sub-agent after each reply
compact	off	Save key insights before context compaction

4. Uninstall

mnemon setup --eject --target openclaw --yes

Workflow

1. Remember: mnemon remember "<fact>" --cat <cat> --imp <1-5> --entities "e1,e2" --source agent
   • Diff is built-in: duplicates skipped, conflicts auto-replaced.
   • Output includes action (added/updated/skipped), semantic_candidates, causal_candidates.
2. Link (evaluate candidates from step 1 — use judgment, not mechanical rules):
   • Review causal_candidates: does a genuine cause-effect relationship exist? causal_signal is regex-based and prone to false positives — only link if the memories are truly causally related.
   • Review semantic_candidates: are these memories meaningfully related? High similarity alone is not sufficient — skip candidates that share keywords but discuss unrelated topics.
   • Syntax: mnemon link <id> <candidate> --type <causal|semantic> --weight <0-1> [--meta '<json>']
3. Recall: mnemon recall "<query>" --limit 10

Commands

mnemon remember "<fact>" --cat <cat> --imp <1-5> --entities "e1,e2" --source agent
mnemon link <id1> <id2> --type <type> --weight <0-1> [--meta '<json>']
mnemon recall "<query>" --limit 10
mnemon search "<query>" --limit 10
mnemon forget <id>
mnemon related <id> --edge causal
mnemon gc --threshold 0.4
mnemon gc --keep <id>
mnemon status
mnemon log
mnemon store list
mnemon store create <name>
mnemon store set <name>
mnemon store remove <name>

Guardrails

• Use the exec tool to run mnemon commands.
• Do not store secrets, passwords, or tokens.
• Categories: preference · decision · insight · fact · context
• Edge types: temporal · semantic · causal · entity
• Max 8,000 chars per insight.