Persistent Mind
PassAudited by ClawScan on May 10, 2026.
Overview
This looks like a local memory tool rather than malware, but anything saved or imported can be reused in future prompts and steer the agent.
PersistentMind appears coherent and local-only based on the provided artifacts. Install it if you want persistent agent memory, but do not store passwords, tokens, or other secrets, and review imported or high-priority memories before letting them shape future prompts.
Findings (2)
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.
Sensitive, outdated, or untrusted memories could be reused later or shared through exports, affecting what the agent sees and does.
The skill intentionally stores memories and reuses them as prompt context, including memories that may be imported from others.
Get a formatted context block to prepend to any prompt ... Export your memory set ... Import a colleague's memories
Avoid storing secrets, review imported memories before use, and periodically delete or archive stale entries.
A bad or stale memory could over-influence the agent in later sessions, even when the current task needs different behavior.
The memory system can elevate stored corrections or high-importance entries into strong future instructions for the agent.
Corrections always surface first. Importance score determines ranking. ... 10 | Critical — never violate
Keep high-importance and correction memories curated, and check the memory context when the agent behaves unexpectedly.