memos-memory-guide
ReviewAudited by ClawScan on May 10, 2026.
Overview
The memory-search purpose is coherent, but the skill also tells the agent to write shared public memory and permanently install or publish skills without clear user confirmation.
Install this only if you want the agent to use long-term conversation memory, and treat public memory writes, skill installation, and skill publishing as actions that should require your explicit review and approval.
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.
The agent may use details from your past conversations and preferences when answering current questions.
The skill is explicitly designed to retrieve and use prior conversation memory, which may include sensitive personal context; this is aligned with its stated purpose.
Search long-term conversation memory for past conversations, user preferences, decisions, and experiences.
Use this only when memory recall is desired, and verify any recalled sensitive details before relying on or sharing them.
Private or sensitive information could be accidentally made available to other agents, and inaccurate public memories could affect future tasks.
The skill instructs use of persistent shared memory visible to all agents, but does not require explicit user confirmation, review, retention limits, or deletion controls before writing.
memory_write_public ... Write a piece of information to public memory. Public memories are visible to all agents during memory_search.
Require explicit user approval before any public memory write, avoid sensitive content, and provide review/delete controls for shared memories.
An unreviewed public skill could become part of the agent's future behavior and influence later sessions.
The guide encourages installing discovered skills, including public skills, into the agent workspace for future use without describing provenance checks or explicit user approval.
Use skill_search to find one first; after finding it, use skill_get to read it, then skill_install to load it for future turns.
Only install skills after user approval and source review; prefer trusted skills and clearly show what will be installed.
A private or sensitive skill could be made discoverable by other agents if the agent publishes it without sufficient review.
Publishing a skill changes public visibility and can propagate content to other agents, but the instructions do not require explicit confirmation before doing so.
skill_publish ... Make a skill public so other agents can discover and install it via skill_search.
Require clear user confirmation before publishing or unpublishing skills, and show exactly what content will become public.