Neural Memory Enhanced
PassAudited by VirusTotal on May 11, 2026.
Overview
Type: OpenClaw Skill Name: neural-memory-enhanced Version: 1.0.0 The skill bundle describes a local associative memory system for AI agents using a neural graph and spreading activation. It provides a comprehensive set of tools for storing and recalling context (e.g., `nmem_remember`, `nmem_recall`) and uses a local SQLite database for storage. While there is a minor metadata inconsistency in `SKILL.md` (listing the installation 'kind' as 'node' while the command is 'pip'), the instructions and logic are consistent with a legitimate utility, and there is no evidence of data exfiltration, malicious execution, or harmful prompt injection.
Findings (0)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Private or outdated conversation details could be saved and later influence the agent’s behavior in unrelated tasks.
The skill directs the agent to persist and later reinject context across sessions, including facts, decisions, errors, and preferences, but the visible instructions do not define user approval, sensitivity exclusions, project boundaries, or retention controls.
Use PROACTIVELY when: ... Starting a new task — inject relevant context from memory ... After making decisions or encountering errors — store for future reference
Use a dedicated brain per project, avoid storing secrets or sensitive personal data, review memory contents regularly, and require explicit user approval before saving or recalling sensitive context.
Sensitive details from normal conversations may be retained longer than the user expects and reused later.
The skill encourages automatic extraction from conversation text into persistent memory. The artifact does not show safeguards for redacting secrets, confirming user intent, or limiting what conversation content may be stored.
At Session End 7. Call `nmem_auto` with action="process" on important conversation segments 8. This auto-extracts facts, decisions, errors, and TODOs
Configure the agent to ask before auto-capturing conversation segments, and establish deletion/redaction practices for stored memories.
A user or installer could fetch a different package source than expected, making it harder to know what code is actually being run.
The install metadata mixes a pip-labeled workflow with a Node install kind, while the setup text tells users to run `pip install neural-memory`. This creates provenance and installation ambiguity for an external package that implements the memory tools.
"install":[{"id":"pip","kind":"node","package":"neural-memory","bins":["nmem"],"label":"pip install neural-memory"}]Verify the intended package repository and package manager before installing, prefer pinned versions, and review the external package if possible.