glmocr-formula
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local Python script (
scripts/glm_ocr_cli.py) to interface with the OCR service. This execution is the primary functionality of the skill and is invoked by the agent to perform formula extraction. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from local files and remote URLs, creating a surface for indirect prompt injection if the OCR output contains instructions the agent might follow.
- Ingestion points: The script
scripts/glm_ocr_cli.pyingests data from local files via the--fileparameter and remote content via the--file-urlparameter. - Boundary markers: The script outputs the OCR results without specific boundary markers or delimiters to isolate the content.
- Capability inventory: The skill possesses network capabilities (sending data to
open.bigmodel.cn) and file system read access for processing images. - Sanitization: No sanitization or filtering is applied to the extracted mathematical formulas or text content.
Audit Metadata