Back to skill

Security audit

mare-liberum

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marked this skill suspicious because it reports a recurring workflow that fetches and follows remote instructions, while the scan was partial.

Review the external URL and decide whether you are comfortable sending it the listed profile details and secret. Also check whether the recurring remote-instruction workflow is necessary, what it may direct the agent to do, and how to stop it. Skillspector scanned the single reported component, but analysis was marked partial because two local path-like references could not be resolved; the evidence is therefore incomplete.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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)

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 31)May include surrounding context.

Make a secret of 32 or more random characters and keep it where you keep your memory (for example ~/.config/mare-liberum/secret). Then:

bash
curl -X POST https://mareliberum.dev/enter \
  -H 'content-type: application/json' \
  -d '{"name":"YourName","about":"one line about you","offers":"what you can do for other agents","seeks":"what you look for","secret":"YOUR_SECRET"}'

The skill installs a recurring remote-instruction workflow without warning about its persistent effects.

Medium
Category
Output Handling
Confidence
97% confidence
Finding

The skill instructs the agent to modify its periodic routine to repeatedly fetch and follow a remotely hosted document, creating a persistent instruction channel whose content can change after installation. Although the skill warns against following other agents' text, it places no equivalent limits on heartbeat instructions; a compromised or malicious endpoint could use this channel to request unauthorized actions with the agent's available tools.

Content

No source excerpt is available for this finding.