Back to skill

Security audit

Wordpress Article Publisher.Skill

Security checks for vulnerabilities and agentic risk

Overview

This WordPress publishing skill is mostly coherent, but it grants live publishing and deletion power with weak safeguards around credentials, HTTPS, and destructive actions.

Review this before installing on a production site. Use a dedicated least-privileged WordPress account, only use HTTPS URLs, prefer draft or staging publication first, avoid passing application passwords in shell commands when possible, and treat any delete or modify-existing-post workflow as destructive unless the target post ID and consequences are explicitly confirmed.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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 (3)

T09 · Insecure Skill Coding Practices

Error
Location
scripts/publish_article.ps1:27
Finding

WordPress credentials can be transmitted over unencrypted HTTP

Content
View full analysis
&1 ``` From `scripts/delete_article.ps1`: ```powershell # Remove the trailing slash from the URL $SiteUrl = $SiteUrl.TrimEnd('/') # Build the API URL $ApiUrl = "$SiteUrl/wp-json/wp/v2/posts/$PostId" # Generate the authentication header $Credentials = "${Username}:${AppPassword}" $EncodedCredentials = [Convert]::ToBase64String([Text.Encoding]::ASCII.GetBytes($Credentials)) Write-Host "Deleting article ID: $PostId ..." # Send a DELETE request with curl $curlArgs = @( "-X", "DELETE", $ApiUrl, "-H", "Authorization: Basic $EncodedCredentials", "-H", "Content-Type: application/json" ) ``` ### Technical Analysis Both scripts accept an arbitrary string as `SiteUrl` and append the WordPress REST API path without parsing the URL or requiring the HTTPS scheme. The resulting request includes an HTTP Basic authentication header containing the Base64-encoded username and application password. Base64 is an encoding mechanism and provides no confidentiality. If a user, compromised workflow, or misleading instruction supplies an `http://` URL, the credentia ...[truncated 1633 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/publish_article.ps1:9
Finding

WordPress application credentials are exposed through process command lines

Content
View full analysis
&1 ``` From `scripts/delete_article.ps1`: ```powershell [Parameter(Mandatory=$true)] [string]$AppPassword, ``` ```powershell $curlArgs = @( "-X", "DELETE", $ApiUrl, "-H", "Authorization: Basic $EncodedCredentials", "-H", "Content-Type: application/json" ) $response = curl.exe @curlArgs 2>&1 ``` The documented invocation in `SKILL.md` also passes the secret directly on the command line: ```powershell .\publish_article.ps1 -SiteUrl "https://3dkee.com" -Username "liuxucai" -AppPassword "xxxx xxxx" -Title "Article title" -Content "

Article content

" ``` ### Technical Analysis The documented interface encourages users to provide the application password as a normal PowerShell command-line argument. Depending on the invocation environment, the plaintext password may be retained in shell history, automation logs, terminal transcripts, endpoint telemetry, or process-creation auditing. The scripts then Base64-encode the username and password and pass the complete `Authorization` header to the external `curl.exe` process as an argument. Although the original password is encoded at this stage, Basic credentials are trivially reversible. Systems or users capable of inspecting process creation data or command-line arguments can decode the captured value. This behavior conflicts with the documentation's general cla ...[truncated 1504 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/publish_article.ps1:57
Finding

Article data is written to a predictable and insufficiently protected temporary file

Content
View full analysis
&1 # Delete the temporary file Remove-Item $jsonFile -Force -ErrorAction SilentlyContinue ``` ### Technical Analysis The script writes the complete article payload to a file under `$env:TEMP`. The name is generated from the current timestamp ticks rather than a cryptographically random value. The file is created using `StreamWriter` without explicit owner-only permissions or an atomic exclusive-creation strategy. The script closes the file before invoking `curl.exe`, creating a period during which another process with access to the relevant temporary directory can read or alter the payload. Whether another unprivileged account can access the file depends on the host's temporary-directory and inherited ACL configuration. Processes running under the same account, administrators, and compromised software in the same security context can generally access it. Cleanup occurs only after `curl.exe` returns and is not placed in a `finally` block. Termination, interruption, or an exception before the cleanup statement can leave ...[truncated 1462 chars]
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (22)

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
80% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · references/api-reference.md (reported line 141)May include surrounding context.

请求:

http
DELETE /wp-json/wp/v2/posts/{id}

可选参数:

Missing User Warnings

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The delete-post section documents force=true as a normal option but does not clearly state that it permanently deletes content and may bypass recovery through the trash. In a publishing automation skill, users or downstream agents could invoke this parameter and cause irreversible content loss.

Content

No source excerpt is available for this finding.

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
93% confidence
Finding

The explicit example DELETE /wp-json/wp/v2/posts/3159?force=true demonstrates a permanently destructive operation without safety guardrails. Because this skill's purpose is to act on a live WordPress site using supplied credentials, such an example can normalize or encourage irreversible deletion flows that are dangerous if triggered accidentally or through prompt manipulation.

Content

Scanner excerpt · references/api-reference.md (reported line 148)May include surrounding context.

  • force=true:永久删除(默认移到回收站)
http
DELETE /wp-json/wp/v2/posts/3159?force=true

分类操作

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

This script adds full post-deletion capability to a skill described as a WordPress article generator/publisher, which materially exceeds the declared scope. In an agent setting, unnecessary destructive actions increase the risk of misuse, prompt/parameter confusion, or unauthorized content removal if the skill is invoked with valid credentials and a post ID.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The code constructs and sends a DELETE request to the WordPress posts API, enabling irreversible destructive behavior that is not justified by the stated publish-only use case. In the skill context, this mismatch is especially dangerous because users may grant credentials expecting content creation, not content deletion, creating a high-risk surprise capability.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
94% confidence
Finding

The skill clearly instructs use of a PowerShell script and shell-based publishing flow, yet the manifest does not declare tool scope or allowed tools. This weakens containment and reviewability because an agent could invoke shell capabilities without explicit user-visible permission boundaries, increasing the chance of unintended command execution or misuse.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The skill asks users to provide WordPress usernames, application passwords, and API endpoints, but does not clearly warn about credential exposure, phishing risk, or the sensitivity of submitting secrets into an agent workflow. Even if the text says passwords are not saved to files, users are still being asked to disclose live publishing credentials that could be mishandled, logged elsewhere, or reused improperly.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill’s documented scope is article generation and publishing, but later expands to deleting and recreating existing posts. That undocumented destructive capability changes the trust model and can lead to data loss or unauthorized content replacement if triggered by a user request or prompt ambiguity.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The documentation claims to 'modify' an existing article, but the implementation actually deletes the original and republishes it. This is dangerous because deletion can break URLs, metadata, revision history, comments, SEO state, or references, while users may believe they are performing a safe in-place edit.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill describes deleting and republishing an article without a clear destructive-action warning or confirmation step. In context, this is more dangerous because the skill operates on a live CMS, so an ambiguous 'modify article' request could cause irreversible content loss, broken links, or publication errors on a production website.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The documentation uses status: "publish" in the primary create-post example without warning that this makes content immediately public. In a skill explicitly designed to generate and publish WordPress articles, this increases the chance of accidental public posting of unreviewed, incorrect, or sensitive AI-generated content.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · references/encoding-fix.md (reported line 32)May include surrounding context.

md
$streamWriter.Write($jsonContent)
$streamWriter.Close()

# 使用 curl 发送
curl.exe -X POST $apiUrl `
    -H "Authorization: Basic $EncodedCredentials" `
    -H "Content-Type: application/json; charset=utf-8" `

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

This markdown file includes a recommended curl.exe example that sends Basic Authorization credentials and post content to a remote WordPress API. The surrounding documentation describes the mechanics of the request but does not warn users that the example transmits credentials and may publish content to a live site, which is a privacy and integrity relevant action under the markdown warning criteria.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The document explicitly instructs users to send their WordPress username and application password to an assistant, but it does not clearly warn users to disclose credentials only to trusted tools, avoid unnecessary sharing, or verify the handling of secrets. Even though application passwords are safer than primary passwords and the broader document includes good security advice, normalizing direct credential sharing with an assistant increases the risk of credential exposure, misuse, or leakage through logs and integrations.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

This code requires a sensitive credential parameter and then builds a Basic Authorization header for an HTTP DELETE request. Although the script logs that it is deleting a post, it does not warn the user that an app password will be used and transmitted to the remote WordPress site, nor does it include any comment or docstring disclosing that sensitive operation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The script defaults the post status to 'publish', which causes an immediate external side effect by creating a live WordPress post without an explicit confirmation step. In a content-generation/publishing skill, this increases the chance of accidental publication of incorrect, harmful, or sensitive content if upstream inputs are wrong or the user did not intend a live post.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The manifest description states the skill is a "WordPress文章发布助手" and describes its behavior entirely in Chinese, indicating a language-specific constraint without any user choice or opt-in. Under the policy rules, forcing a specific language or locale without explicit opt-in is a natural-language policy violation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
82% confidence
Finding

The manifest description is written as a Chinese-only assistant for generating and publishing articles, and the rest of the instructions consistently assume Chinese content and encoding behavior. There is no indication that users may choose another language or that the Chinese-only constraint is an intentional, documented regional limitation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
79% confidence
Finding

The example explicitly writes JSON request content to $env:TEMP\wp-post.json, which can persist user content on the local system. The document does not mention that sensitive draft content may be left in a temp file or suggest cleanup, so users are not warned about a data-handling side effect.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

A language or locale policy issue can arise when a skill forces a specific language without user opt-in. This file presents all instructions exclusively in Chinese and does not indicate that the skill is region-specific or that alternative language support is available.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
96% confidence
Finding

Comments and console output are written entirely in Chinese, including success and failure messages shown to the user. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation unless the locale restriction is explicitly justified.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
98% confidence
Finding

The script’s comments and runtime messages are written in Chinese, including all status and error output. This enforces a specific language/locale experience with no opt-in or configuration, which matches the policy category for language-choice violations.

Content

No source excerpt is available for this finding.