slides-audit

Warn

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill uses the uvx tool to download and install the agent-slides package from an external registry. This package is not pinned to a specific version and does not originate from a recognized trusted organization or well-known service provider.\n- [REMOTE_CODE_EXECUTION]: The execution of uvx --from agent-slides slides triggers the download and immediate execution of code from the external package. This represents a remote code execution vector as the package content is fetched at runtime.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data in the form of PowerPoint files (output.pptx) and design profiles (design-profile.json). Maliciously crafted content within these files could potentially influence the agent's behavior during the analysis or patching steps.\n
  • Ingestion points: output.pptx and design-profile.json files are loaded and parsed in Steps 1, 3, 4, 5, and 6.\n
  • Boundary markers: None identified; the skill relies on the slides tool to process the files without explicit prompt boundaries.\n
  • Capability inventory: The skill has the ability to execute shell commands via uvx and write to the file system using the slides apply subcommand.\n
  • Sanitization: No sanitization or validation of the input presentation content is specified.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Sep 15, 2026, 06:42 AM
Security Audit — agent-trust-hub — slides-audit