The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use shell commands for PDF metadata extraction, text density analysis, and OCR processing. These commands interact with files from the user's filesystem and could be vulnerable to command injection if the filenames or paths are not properly sanitized by the execution environment.
Evidence: OCR.md provides shell script snippets utilizing pdfinfo, pdftotext, ocrmypdf, tesseract, and ghostscript.
[PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it is designed to ingest and act upon content from untrusted external sources (PDFs, scans, emails) placed in an intake folder.
Ingestion points: Data enters the agent's context from files found in ~/Obsidian/Inbox/ as described in SKILL.md.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to distinguish between its system instructions and potential instructions embedded within the ingested documents.
Capability inventory: The skill permits the agent to execute shell commands and perform file write operations to sensitive local directories such as Personal/ and Employment/.
Sanitization: No sanitization, validation, or escaping of the OCR-extracted text is performed before the agent processes the information for classification and reconciliation.

file