slides-grab-image

Warn

Audited by Gen Agent Trust Hub on Aug 8, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
  • [COMMAND_EXECUTION]: The workflow instructs the agent to execute the pdftoppm shell utility using user-provided input paths. This pattern lacks sanitization guidance, which could allow for arbitrary command injection if filenames contain shell metacharacters.
  • [PROMPT_INJECTION]: The skill defines an indirect prompt injection surface. Ingestion points: Content is imported from external PDF or PPTX files during the planning stage. Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the generation prompts. Capability inventory: The skill utilizes shell execution and file system writes. Sanitization: No evidence of validation or filtering for extracted document content is provided.
  • [CREDENTIALS_UNSAFE]: The documentation notes that the skill requires access to sensitive authentication data, including the ~/.codex/auth.json file and various environment variables for API keys such as OPENAI_API_KEY, GOOGLE_API_KEY, and GEMINI_API_KEY.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 8, 2026, 02:07 AM
Security Audit — agent-trust-hub — slides-grab-image