motion-clone

Pass

Audited by Gen Agent Trust Hub on Oct 29, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill downloads necessary binaries during setup, specifically Playwright's Chromium browser and the ffmpeg-static package. These downloads originate from well-known and reputable sources required for the skill's browser automation and video processing functionality.
  • [COMMAND_EXECUTION]: The workflow involves executing local TypeScript automation scripts using the Bun runtime. These scripts perform browser orchestration, image comparison, and video frame extraction as part of the animation cloning process. The setup.sh script also executes package installation commands to prepare the environment.
  • [INDIRECT_PROMPT_INJECTION]: The skill represents an attack surface for indirect prompt injection because it ingests untrusted data from external websites (DOM structures, element metadata, and text content) into the agent's context via inspect.json. This data is subsequently used to generate implementation code.
  • Ingestion points: scripts/capture.ts scrapes DOM content and accessibility trees from user-provided URLs.
  • Boundary markers: The SKILL.md instructions do not define specific delimiters or instructions to ignore embedded commands within the captured website data.
  • Capability inventory: The agent has the capability to write new code files, execute shell commands via bun, and perform network requests via the browser automation tools.
  • Sanitization: No explicit sanitization or filtering is performed on the text content captured from external pages before it is processed by the agent.
  • [DATA_EXFILTRATION]: The capture.ts script includes a mechanism to resolve target URLs to the file:// protocol if no protocol is specified. This allows the skill to theoretically access local files and capture their content (as screenshots or text) if the agent is directed to a local path. However, this is a functional requirement for cloning local prototype exports as described in the skill's documentation.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 29, 2026, 06:21 AM