google-slides-automation

Warn

Audited by Gen Agent Trust Hub on Apr 13, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTIONNO_CODE
Full Analysis
  • [NO_CODE]: The skill references multiple local scripts (scripts/auth.py, scripts/slides.py) that are missing from the provided files. The actual logic for authentication and slide manipulation cannot be audited.
  • [COMMAND_EXECUTION]: Instructs the agent to run local shell commands to perform authentication and slide operations via referenced scripts.
  • [EXTERNAL_DOWNLOADS]: The documentation states that token refreshing is handled via a cloud function. This implies the skill communicates with an external, non-official endpoint for sensitive credential management instead of standard Google OAuth endpoints.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection.
  • Ingestion points: scripts/slides.py get-text (SKILL.md) retrieves arbitrary text from Google Slides presentations.
  • Boundary markers: No delimiters or instructions are provided to the agent to treat slide content as untrusted data.
  • Capability inventory: The skill provides powerful write capabilities including delete-slide, replace-text, and batch-update.
  • Sanitization: There is no evidence of output sanitization or validation of the retrieved text before it enters the agent context.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 13, 2026, 04:38 PM
Security Audit — agent-trust-hub — google-slides-automation