ClawHub
Back to skill

Security audit

Surfagent Perception

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent web-page perception helper, but it needs Review because it handles logged-in pages, screenshots, blockers, and captcha workflows without clear privacy or authorization limits.

Review before installing. Use it only with sites and accounts you are authorized to automate, avoid annotating or sharing pages with passwords, MFA codes, payment data, private messages, or personal records, and protect the SurfAgent daemon token. Treat captcha-solving and blocker automation as sensitive and subject to site rules and legal permission.

A.I.G

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

SkillSpector

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 (3)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The login and form-flow examples normalize typing usernames and passwords into web pages and then verifying state changes, but they do not warn about safe handling of credentials or other form secrets. In practice, this can lead users to paste, log, diff, summarize, or screenshot authentication and form data through the skill pipeline, creating avoidable exposure of secrets and regulated personal data.

Content

No source excerpt is available for this finding.

Ssd 4

Medium
Category
Not specified by scanner
Confidence
81% confidence
Finding

The workflow describes authenticated areas and captchas as ordinary blockers to resolve, and the integration table later includes a captcha-solving action. Even if intended for legitimate automation, this framing can encourage use of the skill to bypass site defenses, access gates, or anti-bot controls without clarifying authorization boundaries, which raises abuse risk.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill explicitly supports annotated screenshots and scene summaries of live web pages, which can capture credentials, personal data, financial details, messages, and other sensitive on-screen content. Because the documentation presents screenshotting and annotation as routine workflow without any privacy warning, redaction guidance, or scope restrictions, it increases the chance that operators will collect and transmit sensitive data unintentionally.

Content

No source excerpt is available for this finding.