Agent Memory Hardened
ReviewAudited by ClawScan on May 3, 2026.
Overview
This appears to be a coherent local memory skill with useful safety guardrails, but users should treat its persistent database as sensitive and verify the install name.
Before installing, verify the package name because the docs mention agent-memory while the reviewed skill is agent-memory-hardened. If you use it, treat ~/.agent-memory/memory.db as sensitive, avoid storing credentials or private personal data, and review recalled memories before the agent takes consequential actions.
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.
Information stored by the agent may reappear in future sessions and influence decisions unless it is reviewed, expired, or deleted.
The skill deliberately stores and recalls cross-session memory from a persistent local database, so stale, incorrect, or overly sensitive memories could affect later agent behavior.
Default: `~/.agent-memory/memory.db` ... On session start: Load recent lessons ... On session end: Extract durable facts from conversation
Use this only for durable, non-sensitive facts; review recalled memory before important actions; configure expirations or periodically delete stale entries.
Following the install snippet without checking the package name could install an unintended or differently reviewed skill.
The skill's declared name differs from the documented install command, which could cause a user to install a different package than the one being reviewed.
name: agent-memory-hardened ... clawdhub install agent-memory
Confirm the exact ClawHub slug and version before installing; prefer the registry entry for agent-memory-hardened if that is the intended skill.