Back to skill

Security audit

Publish Wechat Draft

Security checks for vulnerabilities and agentic risk

Overview

This skill is for a real WeChat publishing workflow, but it asks the agent to run mutable third-party code and use account session cookies with insufficient scoping safeguards.

Install only if you trust the referenced GitHub publisher project and are comfortable letting an agent run npm code that can access your local environment and WeChat publishing session. Prefer pinning the repository and dependencies, running the workflow in a restricted workspace or container, and requiring confirmation before login, installation, file writes, and draft creation.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (2)

T08 · Insecure Dependencies

Error
Location
SKILL.md:36
Finding

Unpinned Third-Party Repository and Dependency Execution

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
SKILL.md:98
Finding

Shell Command Injection Through an Unquoted Configuration Path

Content
View full analysis
``` ### Technical Analysis The Skill directs the Agent to insert a user-provided configuration-file path directly into a Bash command. The placeholder is not quoted, escaped, validated, or passed through a shell-free argument interface. If the Agent performs literal string substitution, shell metacharacters in the supplied path can be interpreted as command syntax rather than as part of a filename. Relevant metacharacters include command separators, command substitutions, redirections, pipes, and shell expansions. For example, a maliciously supplied value conceptually shaped like the following would terminate the intended argument and append another command: ```text config.yaml; attacker-command ``` The exact payload and available commands depend on the host environment. The same unsafe construction pattern can also cause unintended glob expansion, whitespace splitting, or option injection even when an attacker does not achieve full command execution. ### Attack Path 1. An attacker provides or influences a configuration path used for the publishing request. 2. The path contains shell-control syntax rather than only a legitimate filename. 3. The Agent substitutes the untrusted value into the documented Bash command. 4. Bash parses the injected metacharacters as shell syntax. 5. The intended publishing command runs with attacker-supplied trailing operations, or it is replaced or redirected. 6. The injected command executes with the permissions and environment of the Agent process. 7. The attacker can then read or modify accessible files, invoke network tools, interfere with publishing, or execute additional programs. ### Impact Assessment Successful exploitation can provide arbitrary com ...[truncated 677 chars]
Remediation
View remediation
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 (2)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The description includes broad trigger phrases like “发公众号” and especially “发微信,” which can overlap with ordinary conversation and cause the skill to activate in contexts the user did not intend. Because this skill can run shell commands, install dependencies, open browser-driven login flows, and publish content to a real WeChat draft account, accidental invocation could lead to unintended external actions and data handling.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding

The skill instructs use of npx playwright without pinning a specific version, which allows whatever version resolves at execution time to be installed or run. In an agent context that executes shell commands, this creates a supply-chain risk: a compromised upstream package, unexpected major-version change, or registry manipulation could alter behavior or execute unreviewed code on the host.

Content

No source excerpt is available for this finding.