ClawHub
Back to skill

Security audit

Ward Agent Privacy Check

Security checks for vulnerabilities and agentic risk

Overview

This skill is a read-only privacy audit guide that asks agents to map data access and egress while explicitly avoiding secret disclosure and requiring user approval for risky actions.

Before installing, understand that this skill may lead an agent to inspect privacy-relevant configuration and filenames in the scope you name. It is designed to keep checks read-only and redact secrets, but you should still use it with a narrow scope and avoid asking it to inspect home directories, browser profiles, or service accounts unless that is intentional.

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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Credential Access

High
Category
Privilege Escalation
Content
controls before inspecting content.
2. Read configuration and metadata before executing anything. Prefer the
   platform's read-only inspection or help mode.
3. Build a source and capability inventory. A filename such as .env is
   evidence that a candidate exists, not permission to print its contents.
4. Build a sink inventory. Include the model request and logging path when
   their handling is not fully known.
Confidence
60% confidence
Finding
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Credential Access

High
Category
Privilege Escalation
Content
query strings, authorization headers, signed URLs, cookies, and payloads;
- use synthetic markers such as SYNTHETIC_TEST_VALUE in examples;
- do not paste full environment files, browser exports, private keys, session
  transcripts, customer records, or access tokens into prompts, fixtures, or
  reports.

## Observation boundaries
Confidence
70% confidence
Finding
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Vague Triggers

Medium
Confidence
90% confidence
Finding
This is a markdown file, so vague-trigger review applies. The description says to use the skill when a user asks 'what an agent can see, where data may go, whether a skill, plugin, or tool is safe, or how to reduce agent privacy risk,' which is a broad set of natural-language prompts without clear trigger constraints or exclusion conditions, increasing the chance of unintended invocation.

Static analysis

No suspicious patterns detected.