Superlative Memory Manager
Analysis
This instruction-only memory skill is coherent, but it proposes automatic, long-term semantic memory storage without clear boundaries for sensitive data, deletion, or reuse.
Findings (3)
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.
Checks for instructions or behavior that redirect the agent, misuse tools, execute unexpected code, cascade across systems, exploit user trust, or continue outside the intended task.
The skill works automatically once enabled. No manual intervention needed.
Automatic operation is purpose-aligned for a memory manager, but users should notice that compaction, tiering, and recall-related memory handling may occur without per-action confirmation.
"requires": { "skills": [ "cognitive-memory", "memory-tiering", "context-compactor", "memory-on-demand" ] }The submitted skill is instruction-only and depends on other skills for actual memory behavior. Those dependencies are disclosed, but their code and permissions are outside this artifact set.
Checks for exposed credentials, poisoned memory or context, unclear communication boundaries, or sensitive data that could leave the user's control.
Tiered Storage: Hot (session), Warm (LanceDB vectors), Cold (Git-notes), Archive (compressed)
The skill proposes retaining agent memory across multiple persistent stores, including vector indexes, Git notes, and archives. The artifact does not clearly define storage boundaries, exclusions for sensitive data, deletion controls, or reuse limits.