ClawHub
Back to skill

Security audit

AutoSignals

Security checks for vulnerabilities and agentic risk

Overview

This skill appears to run an autonomous trading-research loop that edits local files and writes logs continuously, which is not clearly unsafe but needs review before use.

Install only if you intend to run autonomous trading-signal experiments, preferably in a dedicated sandbox or disposable repository. Review the affected files, logging location, start/stop commands, and resource impact before enabling the loop, and avoid connecting it to live trading or sensitive credentials unless you have separately verified the surrounding controls.

SkillSpector

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

Missing User Warnings

Medium
Confidence
91% confidence
Finding
The README explicitly describes a continuously running autonomous loop that spawns agents to modify `signals.py` and persist experiment logs, but it does not prominently warn users that enabling the skill will cause ongoing file modifications and continuous disk writes. In an agent skill context, this can lead to unintended local code changes, resource consumption, and sensitive experiment artifacts being retained without clear user consent or operational safeguards.

Vague Triggers

Medium
Confidence
88% confidence
Finding
The skill has no trigger patterns or activation constraints, yet it exposes operational commands for a continuously running autonomous trading loop. In an agent environment, vague activation boundaries increase the chance the skill is invoked in unintended contexts, causing the agent to surface or execute high-impact operational guidance without clear user intent.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The skill includes a direct process-termination command (`kill $(cat .../autosignals.pid)`) without any warning, validation, or confirmation step. Even though it targets a PID file, it still instructs destructive action against a running process; if invoked carelessly, or if the PID file is stale or tampered with, it could terminate the wrong process and disrupt trading research operations.

Static analysis

No suspicious patterns detected.