ClawHub
Back to skill

Security audit

dawn-memory-arch

Security checks for vulnerabilities and agentic risk

Overview

This is a local long-term memory architecture skill, but it automatically persists profile, preference, session, vector-search, and trading/account-related data without clear user controls.

Review this carefully before installing if you do not want an agent to retain conversation details, preferences, profile facts, session summaries, semantic-search entries, or trading/account state across sessions. It appears local and not exfiltrative, but you should add or require explicit consent, redaction, review, deletion, and retention controls before using it with sensitive information.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Vague Triggers

Medium
Confidence
85% confidence
Finding
The skill defines broad automatic triggers such as corrections, decisions, preference changes, and value changes without clear scope, authorization boundaries, or validation rules. In a memory-management skill that writes to local files, this can cause excessive or unintended persistence of sensitive, low-quality, or adversarially injected content, degrading integrity of long-term memory and influencing later agent behavior.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill instructs automatic writes to local artifacts such as session-state.json, MEMORY.md, working-buffer.md, state.db, and the vector database, but does not prominently warn the user that interaction content may be persisted automatically. This creates a transparency and privacy risk because users may provide sensitive data without realizing it will be stored across sessions and potentially reused in future decisions or retrieval flows.

Static analysis

No suspicious patterns detected.