ClawHub
Back to skill

Security audit

geo-proposal

Security checks across malware telemetry and agentic risk

Overview

This skill does proposal generation as advertised, but it can be invoked too broadly and can change local business records without an explicit confirmation step.

Install only if you are comfortable with the agent creating proposal files and changing prospect status records in ~/.geo-prospects. Prefer invoking it explicitly with /geo proposal and review target paths before allowing writes, especially in workspaces containing real client or sales data.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Vague Triggers

Medium
Confidence
93% confidence
Finding
The skill metadata advertises activation on broad, common phrases such as 'proposal', 'proposta', 'offerta', and 'preventivo'. This can cause unintended invocation in unrelated conversations, leading the agent to read audit files, generate sales documents, or trigger follow-on actions without the user explicitly requesting this skill. Because the skill also has write capabilities and persistent side effects, accidental triggering is more dangerous in this context than for a read-only skill.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The workflow instructs the skill to write a proposal file and update prospect records, but it does not require a user warning or confirmation before making persistent changes. This creates a risk of silent state modification in local files, including changing prospect status and recording paths, which can surprise users, corrupt workflow state, or leak sensitive business data into unintended locations. The presence of Write and Bash permissions increases the risk because the skill is operationally capable of making those changes.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.