ClawHub
Back to skill

Security audit

SwarmRecall Dream

Security checks for vulnerabilities and agentic risk

Overview

This skill does what it says: it connects to SwarmRecall to clean up and summarize agent memory, with disclosed remote read/write effects.

Install only if you trust SwarmRecall with your agent's stored memories. Use an API key you control, review auto-dream settings before enabling them, and prefer explicit confirmation before running pruning, merging, or contradiction-resolution operations.

SkillSpector

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Context-Inappropriate Capability

Medium
Confidence
93% confidence
Finding
The skill instructs the agent to self-register for a new external service account and persist the returned API key into the environment, which is a capability expansion beyond simple memory-maintenance behavior. This can cause unapproved account creation, secret acquisition/storage, and silent outbound data onboarding to a third-party service without an explicit user action or administrator-approved provisioning flow.

Context-Inappropriate Capability

Medium
Confidence
74% confidence
Finding
The unprocessed-memories endpoint is described as being for knowledge-graph enrichment, which extends the skill into broader data transformation and inference beyond the stated dreaming scope of consolidation, pruning, deduplication, contradiction resolution, and summarization. Scope creep like this increases the chance that the agent will process or derive additional structured data without clear user expectation or least-privilege boundaries.

Vague Triggers

Medium
Confidence
84% confidence
Finding
The explicit-request trigger examples such as 'clean up your memory' or 'consolidate what you know' are broad natural-language phrases that may be uttered casually and interpreted as authorization to perform destructive or privacy-impacting remote modifications. In this skill, those actions include archival, pruning, and contradiction resolution against server-hosted memory, so ambiguous triggers materially raise the risk of unintended state changes.

Static analysis

No suspicious patterns detected.