Three-Dimensional Memory

ReviewAudited by ClawScan on May 10, 2026.

Overview

The skill is mostly coherent for a memory system, but it encourages persistent full conversation backups and long-term agent preference/personality files without clear consent, retention, or redaction controls.

Install only if you want the assistant to maintain local memory files. Before using it, decide which conversations may be saved, where the files should live, how long they should be kept, and how to remove or redact sensitive information.

Findings (4)

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.

What this means

Private conversation content may be saved in local files and later searched, copied, synced, or reused beyond the original chat context.

Why it was flagged

The skill explicitly directs preservation of complete conversation transcripts, which can contain sensitive personal, business, or credential-like information.

Skill content
Content: Every word exchanged, including user messages and AI responses
Recommendation

Use this only in a trusted workspace, require explicit user approval before saving full transcripts, redact secrets, and define retention and deletion rules.

What this means

If these files are inaccurate, stale, or edited by mistake, future agent behavior could be influenced by untrusted or unwanted persistent context.

Why it was flagged

The skill recommends persistent files that can shape the agent's future behavior and store user preferences, but does not define trust boundaries or validation for those files.

Skill content
MEMORY.md ← AI's long-term identity; SOUL.md ← AI's personality; USER.md ← User preferences
Recommendation

Keep persistent memory files separate from instructions, review changes manually, and avoid storing secrets or authoritative commands in memory files.

What this means

Running the example in the wrong directory or rerunning it could create or replace local memory files.

Why it was flagged

The example workflow writes a daily memory file and would overwrite the same day's file if run again.

Skill content
cat > "memory/${TODAY}.md" << EOF
Recommendation

Run examples only in the intended workspace and consider changing overwriting writes to append or prompt before replacing existing files.

What this means

Users may overestimate the reliability or completeness of the skill based on marketing language.

Why it was flagged

The README uses a strong guarantee and also advertises command examples that are not implemented in the provided files.

Skill content
Find any file in 10 seconds. Guaranteed.
Recommendation

Treat the README as a workflow concept rather than a guarantee, and verify any advertised commands exist before relying on them.