Back to skill

Security audit

waf-validator-skills

Security checks for vulnerabilities and agentic risk

Overview

This is a coherent WAF testing skill, but it needs review because it guides active malicious-request testing and unpinned host-network Docker execution without enough scoping or authorization safeguards.

Install only if you understand that this skill is for authorized WAF/security validation and may send malicious-looking requests to live systems. Prefer staging targets, written authorization, pinned container digests, restricted Docker networking, and a dedicated empty reports directory.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (1)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:155
Finding

Unpinned External Container Executed with Host Networking

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 155–157
Vulnerability Type: Unpinned third-party container dependency with excessive network access
Risk Level: Medium

Vulnerable Code

bash
docker pull wallarm/gotestwaf
docker run --rm --network="host" -v ${PWD}/reports:/app/reports \
    wallarm/gotestwaf --url=http://target-url --noEmailReport

Technical Analysis

The documented workflow pulls and executes wallarm/gotestwaf without specifying an immutable version or image digest. The effective executable payload can therefore change after this skill has been reviewed. A tag update, registry compromise, or publisher-account compromise could cause users to execute different code than expected.

The container is also granted host-network access through --network="host" and write access to ${PWD}/reports. Host networking removes Docker's normal network namespace isolation, allowing processes inside the container to communicate directly with services reachable through the host network stack. The bind mount exposes the selected local report directory to container modification.

The audited artifact contains only SKILL.md; it does not include the source tree referenced by the build instructions. Consequently, the external image's implementation cannot be verified against local source in this artifact.

Attack Path

  1. An attacker compromises the image publisher, registry account, build pipeline, or mutable image tag.
  2. The attacker publishes a modified wallarm/gotestwaf image under the referenced tag.
  3. A user follows the documented command and pulls the modified image.
  4. Docker executes the attacker-controlled image with host networking enabled.
  5. The malicious process enumerates or accesses host-reachable services and other network resources available from the host.
  6. The process reads or alters content available through the mounted ${PWD}/reports directory.

Imp

...[truncated 740 chars]

Remediation
View remediation

Remediation Suggestions

  1. Pin the reviewed container to an immutable digest:
    bash
    docker pull wallarm/gotestwaf@sha256:<verified-digest>
    docker run --rm \
      --network=<restricted-network> \
      --mount type=bind,src="${PWD}/reports",dst=/app/reports \
      wallarm/gotestwaf@sha256:<verified-digest> \
      --url=http://target-url --noEmailReport
    
  2. Verify the digest through a trusted release channel and validate available image signatures or attestations before execution.
  3. Replace --network="host" with a dedicated, restricted Docker network. Permit access only to the authorized WAF test target.
  4. Mount only a dedicated output directory with narrowly scoped host permissions. Do not expose unrelated or sensitive files.
  5. Run the container as a non-root user, drop unnecessary Linux capabilities, enable no-new-privileges, and use a read-only root filesystem where compatible:
    bash
    --user <non-root-uid> \
    --cap-drop=ALL \
    --security-opt=no-new-privileges \
    --read-only
    
  6. Include the corresponding reviewed source code and reproducible build instructions in the package so the image can be independently verified.
Vulnerability Patterns
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (5)

Vague Triggers

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The manifest uses very broad single-word triggers like waf, rule, security, and testing, which can cause the skill to activate in many unrelated conversations. In this skill's context, accidental activation is more dangerous than usual because the content provides guidance for generating malicious requests and actively probing target systems.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill describes offensive testing functionality but does not prominently warn that it generates malicious payloads and scans live targets. Without a clear safety and authorization warning, users may run it against systems they do not own or in production environments, creating legal, operational, and security risk.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The main description line is written in Chinese while the rest of the document is in English, and there is no indication that the skill is intentionally region-specific or that users can choose their preferred language. This creates a language-policy issue because the skill implicitly imposes a locale without opt-in.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding

The skill instructs users to pull a Docker image by mutable name only, without a fixed tag or digest. That allows upstream image changes or a compromised registry namespace to silently alter what gets executed, which is especially risky here because the tool generates active attack traffic and will often be run with host networking.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
75% confidence
Finding

Docker image references without a specific tag (:latest is implicit) or digest (@sha256:...) can be silently replaced by a malicious image.

Content

No source excerpt is available for this finding.