full-width-artifact

Pass

Audited by Gen Agent Trust Hub on Nov 7, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The script executes external binaries including pandoc for Markdown rendering and the orca CLI for publishing artifacts. It includes a discovery mechanism for the Orca CLI that validates the binary's help output before execution to ensure it is the expected tool.
  • [EXTERNAL_DOWNLOADS]: The verification step in the script uses the fetch API to retrieve the published artifact from share.onorca.dev and its embedded iframes. This is used solely to verify that the layout is rendered correctly after publishing.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external Markdown and HTML files and retrieves content from remote URLs for layout verification, which represents a potential attack surface for indirect instructions.
  • Ingestion points: The orca-artifact-share script reads local files via readFileSync and fetches remote content via fetch during the verification phase in verifyRemote.
  • Boundary markers: No specific delimiters or safety warnings are used when processing the input file content or the fetched remote content.
  • Capability inventory: The script has the capability to execute subprocesses (spawnSync), write to the local file system (specifically in ~/.local/share/orca-artifact), and perform network requests (fetch).
  • Sanitization: The script performs layout-specific transformations and escapes the document title, but it does not specifically sanitize the content for prompt injection patterns. The transformations are primarily focused on CSS properties and layout markers.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 7, 2026, 02:12 AM