ClawHub
Back to skill

Security audit

WBFlow·批量工厂总控（调度层）

Security checks for vulnerabilities and agentic risk

Overview

The video-batching workflow is coherent, but unsafe file handling creates risks of overwriting unrelated files or including unintended media.

Use trusted batch configurations and manifests. Before enabling montage, address temporary-file protection, filename escaping, and directory containment. Review the companion skills before enabling scheduled production. These concrete implementation concerns warrant review, but the artifacts do not establish malicious intent.

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
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • 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 (1)

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
95% confidence
Finding

Running ffmpeg with an argument list and no shell is appropriate for this video-batching skill and does not itself introduce shell-command injection. However, its input list is written to a predictable shared temporary path using an open operation that follows symlinks, allowing a local attacker to cause truncation of another file writable by the process or tamper with the list before ffmpeg reads it. Additionally, filenames are interpolated into the concat format without escaping, so attacker-controlled filenames containing quotes and newlines can inject directives that include unintended local media, especially with '-safe 0'.

Content

Scanner excerpt · scripts/aggregate.py (reported line 108)May include surrounding context.

python
"-i", concat, "-c", "copy", out]
    print(f"\n拼接合集：{' | '.join(os.path.basename(p) for p in finals)}")
    try:
        subprocess.run(cmd, check=True, capture_output=True, text=True)
        print(f"已生成合集预览：{out}")
    except FileNotFoundError:
        print("[ERROR] 未找到 ffmpeg（需置于 PATH），合集跳过", file=sys.stderr)