Back to skill

Security audit

Meitu Beauty

Security checks for vulnerabilities and agentic risk

Overview

The skill appears purpose-aligned for Meitu photo beautification, but it asks for sensitive credential access and gives unsafe execution/install instructions around user photos.

Review before installing. Use this only if you are comfortable sending portrait photos to Meitu, prefer environment or managed secrets over granting direct access to ~/.meitu/credentials.json, pin and verify the CLI version, and avoid running the documented shell templates with untrusted filenames or URLs unless they are safely escaped or passed as structured arguments.

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

Warning
Location
SKILL.md:33
Finding

Unpinned Global Installation of a Credential-Handling Dependency

Content
View full analysis
**路径别名:** 下文中 `$VISUAL` = `{OPENCLAW_HOME}/workspace/visual/` ## Core Workflow ... ### Preflight 1. `meitu --version` → 未安装则提示 `npm install -g meitu-cli` ``` ### Technical Analysis The Skill instructs users to install `meitu-cli` globally from the npm registry without specifying an exact reviewed version, integrity hash, lockfile, or verified distribution source. Although the documentation states that version 0.1.9 or later is required, the command does not enforce that constraint and normally resolves to the registry's current release. The installed CLI is subsequently entrusted with both Meitu API credentials and user-supplied portrait images. A compromised package release, compromised publisher account, or malicious dependency introduced into a later release could therefore execute under the user's account. npm packages may also run lifecycle scripts during installation, increasing exposure before the CLI is explicitly invoked. The audit did not find evidence that the named package is currently malicious. The vulnerability is the unsafe, mutable installation process and the level of trust granted to the resulting executable. ### Attack Path 1. An attacker compromises the npm package, its publisher account, or a transitive dependency used by a later release. 2. A user follows the Skill's instruction and runs `npm install -g meitu-cli`. 3. npm resolves and installs the attacker-controlled release because no exact version or integrity value is enforced. 4. Malicious code executes through an installation life ...[truncated 798 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
SKILL.md:52
Finding

Unsafe Shell Interpolation of Paths, URLs, and Filenames

Content
View full analysis
\ --beatify_type <0|1> \ --json \ --download-dir {output_dir} ``` ```markdown `mv {file} {output_dir}/{date}_beauty_{original_name}.{ext}` ``` ### Technical Analysis The documented command templates interpolate multiple values into shell commands without requiring shell-safe quoting or argument-array execution: - `output_dir` may be derived from `openclaw.yaml`. - `url_or_path` originates from a user-provided URL or local path. - `file` is obtained from tool output. - `original_name` is derived from the supplied image filename. - `ext` may also originate from the input filename or media metadata. If an agent materializes these templates as raw shell commands, whitespace and shell metacharacters in substituted values may be interpreted by the shell rather than treated as literal data. Crafted values containing command separators, substitutions, redirections, glob characters, or newlines could consequently alter the executed command. Leading hyphens also create an option-injection risk for commands such as `mv` unless the end-of-options marker is used. Even without arbitrary command execution, unquoted whitespace or wildcard expansion can cause operations to affect unintended files. The exploitability is conditional on these templates being executed through a shell with unescaped substitutions. Argument-array process APIs would prevent shell metacharacter interpretation, but the Skill does not require that safer execution method. ### Attack Path 1. An attacker supplies a portrait with a crafted local filename, image URL, or project configuration value containing shell syntax or a leading option. 2. The Skill derives `url_or_path`, `output_dir`, or `o ...[truncated 1287 chars]
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Rogue AgentSelf-Modification, Session Persistence
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (7)

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 5)May include surrounding context.

md
name: meitu-beauty
description: "对人像照片进行 AI 美颜处理(磨皮、美白、精修五官)。当用户提到美颜、磨皮、美白、精修、beautify、beauty enhance、让照片更好看时触发。仅支持单人照片。"
version: "1.1.0"
metadata: {"openclaw":{"requires":{"bins":["meitu"],"env":["MEITU_OPENAPI_ACCESS_KEY","MEITU_OPENAPI_SECRET_KEY"],"paths":{"read":["~/.meitu/credentials.json","~/.openclaw/workspace/visual/"],"write":["~/.openclaw/workspace/visual/"]}},"primaryEnv":"MEITU_OPENAPI_ACCESS_KEY"}}
requirements:
  credentials:
    - name: MEITU_OPENAPI_ACCESS_KEY

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 9)May include surrounding context.

md
name: meitu-beauty
description: "对人像照片进行 AI 美颜处理(磨皮、美白、精修五官)。当用户提到美颜、磨皮、美白、精修、beautify、beauty enhance、让照片更好看时触发。仅支持单人照片。"
version: "1.1.0"
metadata: {"openclaw":{"requires":{"bins":["meitu"],"env":["MEITU_OPENAPI_ACCESS_KEY","MEITU_OPENAPI_SECRET_KEY"],"paths":{"read":["~/.meitu/credentials.json","~/.openclaw/workspace/visual/"],"write":["~/.openclaw/workspace/visual/"]}},"primaryEnv":"MEITU_OPENAPI_ACCESS_KEY"}}
requirements:
  credentials:
    - name: MEITU_OPENAPI_ACCESS_KEY

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 11)May include surrounding context.

md
name: meitu-beauty
description: "对人像照片进行 AI 美颜处理(磨皮、美白、精修五官)。当用户提到美颜、磨皮、美白、精修、beautify、beauty enhance、让照片更好看时触发。仅支持单人照片。"
version: "1.1.0"
metadata: {"openclaw":{"requires":{"bins":["meitu"],"env":["MEITU_OPENAPI_ACCESS_KEY","MEITU_OPENAPI_SECRET_KEY"],"paths":{"read":["~/.meitu/credentials.json","~/.openclaw/workspace/visual/"],"write":["~/.openclaw/workspace/visual/"]}},"primaryEnv":"MEITU_OPENAPI_ACCESS_KEY"}}
requirements:
  credentials:
    - name: MEITU_OPENAPI_ACCESS_KEY

Credential Access

High
Category
Privilege Escalation
Confidence
72% confidence
Finding

Granting explicit file_read permission to ~/.meitu/credentials.json gives the skill direct access to stored API credentials. Even though the stated purpose is legitimate, unnecessary direct secret-file access increases the blast radius if the skill is modified, compromised, or mis-triggered, because the agent could read reusable secrets from disk.

Content

Scanner excerpt · SKILL.md (reported line 15)May include surrounding context.

md
permissions:
    - type: file_read
      paths:
        - ~/.meitu/credentials.json
        - ~/.openclaw/workspace/visual/
    - type: file_write
      paths:

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The trigger text includes broad everyday phrases like '让照片更好看' and generic beautify terms, which can cause the skill to activate in situations where the user did not explicitly consent to using this external image-processing service. Because the skill handles user photos and may send them to a third-party API, accidental activation increases privacy and data-handling risk.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · SKILL.md (reported line 17)May include surrounding context.

md
paths:
        - ~/.meitu/credentials.json
        - ~/.openclaw/workspace/visual/
    - type: file_write
      paths:
        - ~/.openclaw/workspace/visual/
    - type: exec

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill description and workflow do not clearly disclose that local image paths, uploaded photos, or image URLs will be sent to Meitu's external beautification service. This can lead to users unintentionally sharing sensitive biometric or personal imagery with a third party without informed consent.

Content

No source excerpt is available for this finding.