reading-deduction-cert
Pass
Audited by Gen Agent Trust Hub on Feb 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes a CLI tool named
shinkokuto process PDF files. - Evidence:
shinkoku pdf extract-text --file-path <path>andshinkoku pdf to-image --file-path <path> --output-dir <dir>are executed inSKILL.md. - The commands are used for the primary purpose of the skill (PDF processing) and target specific utility functions.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted external data in the form of images and PDF files.
- Ingestion points: The skill reads
.pdfand image files (PNG) provided by the user (found inSKILL.md). - Boundary markers: No explicit delimiters or instructions to ignore embedded text are used when passing the image content to the sub-agents.
- Capability inventory: The skill has the ability to execute shell commands via
shinkokuand spawn sub-agents. - Sanitization: No evidence of file path sanitization or image content filtering is present in the static instructions.
Audit Metadata