Back to skill

Security audit

Setup Doctor

Security checks for vulnerabilities and agentic risk

Overview

Setup Doctor is a disclosed local diagnostics skill that checks tool versions, gateway status, and OpenClaw file presence without evidence of credential theft, persistence, or hidden modification.

Install only if you are comfortable with a troubleshooting skill running local version and OpenClaw status checks when you report setup errors. Review proposed fixes before confirming them, and treat the full-suite install suggestion as optional because it installs additional skills.

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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (8)

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The skill claims restrictive behavior such as never reading file contents and only checking existence/size, yet the analyzer indicates implemented behavior may test readability/access and diverge from the documented modes and outputs. This kind of description-behavior mismatch is dangerous because users may consent based on a narrower privacy model than what the skill actually does, enabling overcollection or unintended actions under false expectations.

Content

No source excerpt is available for this finding.

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · SKILL.md (reported line 114)May include surrounding context.

md
## What This Skill Does NOT Do

- Does NOT read file contents (existence/size only)
- Does NOT access credentials without explicit approval
- Does NOT modify config automatically
- Does NOT modify HEARTBEAT.md or MEMORY.md

Undeclared Tool Scope

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.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 3)May include surrounding context.

md
---
name: setup-doctor
description: "Diagnose OpenClaw setup in one command. Auto-detects issues with Node, npm, gateway, config, and workspace. Quick mode (~10s) or Full mode. Multi-language. Config and workspace checks report file existence only — never reads contents. No credentials accessed without approval. Homepage: https://clawhub.ai/skills/setup-doctor"
---

# Setup Doctor v3.0

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The auto-run triggers are broad enough to activate on ordinary troubleshooting phrases like 'doesn't work' or generic error reports, which can cause shell commands and environment inspection without a clear, specific request. In an agent setting, overbroad invocation increases the chance of unintended command execution, surprise data exposure about the system state, and user confusion about what the agent is doing autonomously.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 67)May include surrounding context.

md
3. Apply fix
4. Re-run check to verify

**Never fix without confirmation.**

## Workspace Audit

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 108)May include surrounding context.

md
2. **Keep it fast** — quick check in ~10 seconds
3. **One-line fixes** — clear, copy-pasteable solutions
4. **Never read file contents** — existence and size only
5. **Never fix without asking** — show what changes, get confirmation
6. **Match user language** in reports

## What This Skill Does NOT Do

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The remediation message at L113 tells the user to 'Fix JSON syntax', implying the script validated configuration contents. In reality, the code only checks for file existence and read access via fs.accessSync and never parses or reads the config file, so the documentation overstates what the check actually determined.

Content

No source excerpt is available for this finding.