ClawHub
Back to skill

Security audit

Tmp.U6ndBhryVk

Security checks across malware telemetry and agentic risk

Overview

This skill is transparent about controlling Evite from the shell, but it gives broad live-account write access through private endpoints without strong safeguards.

Install only if you are comfortable letting an agent use your Evite credentials and live session cookies to read guest data and perform real event actions. Keep write operations user-approved, test with throwaway events first, avoid printing or storing guest emails/phone numbers unnecessarily, and confirm exact file paths and upload destinations before photo uploads.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (5)

Missing User Warnings

Medium
Confidence
81% confidence
Finding
The skill includes concrete state-changing Evite operations and explicitly notes that writes will mutate real events, RSVPs, cancellations, and guest communications, but the warning appears late in the document rather than before the actionable write examples. In an agent skill context, this increases the risk of unintended real-world actions because a user or downstream agent may follow the examples before encountering the cautionary note.

Natural-Language Policy Violations

Medium
Confidence
87% confidence
Finding
The skill is designed to authenticate against Evite's internal, non-public /services/, /ajax/, and /tsunami/ endpoints rather than a documented public API. Using private interfaces is dangerous because they may bypass intended integration controls, are unstable, and can expose users to account lockouts, terms-of-service violations, or unexpected security behavior when automated by an agent.

Missing User Warnings

Medium
Confidence
89% confidence
Finding
The documented guest-list endpoint returns personal contact data including guest names, email addresses, and phone numbers, but the reference does not warn operators that they are accessing third-party PII. In a shell-oriented skill, this increases the chance that sensitive attendee data will be fetched, displayed, logged, or reused without informed user consent or data-minimization safeguards.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The photo-upload instructions state that files are sent to Google Cloud Storage rather than directly to Evite, but they do not provide a clear user-facing warning that content will be transferred to a third-party service. This can cause users to unknowingly upload personal images and metadata to external infrastructure, which is especially sensitive in a shell automation context.

External Transmission

Medium
Category
Data Exfiltration
Content
v=$(jq -r --arg k "$k" '.upload_form[$k]' <<<"$TICKET")
  FORM_ARGS+=(-F "$k=$v")
done
FINISH_URL=$(curl -sS -D - -o /dev/null "${FORM_ARGS[@]}" -F "file=@/path/to/photo.jpg" "$UPLOAD_URL" \
  | grep -i '^location:' | tr -d '\r' | awk '{print $2}')

# Step 3 — finalize the object into the album (best-effort; ignore failures).
Confidence
91% confidence
Finding
curl -sS -D

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.