Back to skill

Security audit

网页后台上传自动化(免安装浏览器)

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious because its browser-debugging guidance raises concerns about access to signed-in browser sessions and changes to drafts; the review is medium confidence and the scan is incomplete.

Consider avoiding installation unless you need this workflow and can use an isolated browser profile with only the necessary account session. Review the WebSocket-origin guidance and draft-saving steps carefully; scanner coverage is partial, with some local references unresolved and a bundled image outside analyzer coverage.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Taint TrackingDirect Taint Flow, Variable-Mediated Taint Flow, Credential Exfiltration Chain
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (6)

Tainted flow: 'req' from os.environ.get (line 58, credential/environment) → urllib.request.urlopen (network output)

Critical
Category
Data Flow
Confidence
90% confidence
Finding

Credentials or environment variables flow to a network sink. This is a high-confidence indicator of credential exfiltration.

Content

Scanner excerpt · scripts/cdp.py (reported line 59)May include surrounding context.

python
# ---------------- CDP ----------------
def targets():
    req = urllib.request.Request('<redacted-url>' % PORT)
    return json.load(urllib.request.urlopen(req, timeout=10))


def pick():

The browser-debugging setup omits a warning about access to authenticated browser sessions.

High
Category
Output Handling
Confidence
97% confidence
Finding

The setup exposes a powerful browser-debugging interface and explicitly suggests allowing every WebSocket origin, weakening protection against connections from malicious web pages. Because the workflow uses authenticated browser sessions, unauthorized CDP access could expose account data or perform actions as the user; recommending a temporary profile and eventual shutdown reduces but does not eliminate that risk.

Content

No source excerpt is available for this finding.

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: env, network.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

The style-probing workflow saves experimental content without warning that it can overwrite the current draft.

Medium
Category
Output Handling
Confidence
95% confidence
Finding

The probe workflow directs the agent to inject experimental HTML and save it without requiring a disposable draft or a recoverable backup. In an authenticated content-management session, applying these steps to the current document could persistently overwrite valuable content, even when the user only requested a styling diagnosis.

Content

No source excerpt is available for this finding.

Internal Network Request

Medium
Category
Server-Side Request Forgery
Confidence
70% confidence
Finding

Code issues a request to a loopback, link-local, or private-range host. This can reach internal services not meant to be exposed and is a common SSRF pivot.

Content

Scanner excerpt · scripts/cdp.py (reported line 58)May include surrounding context.

python
# ---------------- CDP ----------------
def targets():
    req = urllib.request.Request('<redacted-url>' % PORT)
    return json.load(urllib.request.urlopen(req, timeout=10))

The final activation phrase covers generic UI failures without requiring a web-console automation context.

Low
Category
Trigger Abuse
Confidence
90% confidence
Finding

The trigger “点了按钮没反应/提交没成功也没报错” means a button did nothing or submission failed silently, which can describe unrelated websites and desktop applications. Unlike the other triggers, it does not require an authenticated web console, an upload workflow, or an existing CDP automation task.

Content

No source excerpt is available for this finding.