Security audit

视频提取

Security checks for vulnerabilities and agentic risk

Overview

This skill is a narrow ffmpeg helper for extracting video frames, with one input-validation issue users should handle carefully.

Install only if you are comfortable using ffmpeg on local video files. Use explicit output paths you are willing to overwrite, and avoid passing untrusted text as --index until the script validates it as a non-negative integer.

A.I.G

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 (1)

T09 · Insecure Skill Coding Practices

Warning

Location: scripts/frame.sh:61
Finding: FFmpeg Filtergraph Injection Through Unvalidated Frame Index
Content: View full analysis
Remediation: View remediation
&2 exit 2 fi ``` Place this validation after option parsing and before `mkdir` or any FFmpeg invocation. Additional hardening measures: 1. Reject missing values explicitly for `--index`, `--time`, and `--out` rather than allowing a subsequent option to be consumed as a value. 2. Apply a reasonable maximum frame index to reduce accidental or intentional resource consumption. 3. Validate `--time` against a strict supported timestamp format. 4. Continue passing all filenames as quoted arguments. 5. Add negative tests covering filtergraph metacharacters, alphabetic values, negative numbers, empty values, and excessively large integers. 6. Run FFmpeg with least privilege and restrict filesystem and network access when processing untrusted media. ]]>

SkillSpector

Vulnerability Patterns

MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
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)

Tp4

High

Category: MCP Tool Poisoning
Confidence: 95% confidence
Finding: The script clearly supports frame extraction from a video by timestamp, by frame index, or by defaulting to the first frame. However, it does not implement short clip extraction at all, so the declared description overstates its capabilities. It also creates output directories and writes the extracted frame to a specified path, which is filesystem write access not reflected in the declared permissions list. There are no suspicious unrelated behaviors beyond the video/frame extraction workflow.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.