extract-text-from-image
Warn
Audited by Gen Agent Trust Hub on Apr 12, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses shell commands like tesseract and convert to process user-provided image and PDF files. The instructions interpolate user-controlled variables directly into command strings, which can lead to command injection if the agent does not properly validate or escape the file paths.
- [EXTERNAL_DOWNLOADS]: The skill performs runtime installation of system packages such as tesseract-ocr, imagemagick, and ghostscript using apt-get. This involves downloading software from external repositories at execution time.
- [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by converting untrusted visual data into text that the agent subsequently reads.
- Ingestion points: Extracted text from images and PDFs is written to files like output.txt and combined_output.txt as shown in SKILL.md.
- Boundary markers: Absent; there are no instructions provided to the agent to isolate the extracted content or treat it as untrusted data.
- Capability inventory: The skill has the ability to execute shell commands and modify the system via package installation.
- Sanitization: Absent; the extracted text is read directly into the agent context using cat.
Audit Metadata