The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted PDF documents, which serves as a vector for indirect prompt injection attacks.
Ingestion points: The skill loads PDF files for text and table extraction using pdfplumber and pytesseract as outlined in SKILL.md and instructions.md.
Boundary markers: There are no explicit instructions or delimiters defined to help the agent distinguish between legitimate document content and potential malicious instructions embedded within the PDF data.
Capability inventory: The skill utilizes filesystem read/write permissions and has the capability to execute Python scripts that generate structured output files (Excel, CSV, JSON).
Sanitization: While SKILL.md includes a clean_extracted_data function, it focuses on data types and whitespace rather than filtering for adversarial prompt instructions.
[EXTERNAL_DOWNLOADS]: The skill documentation instructs the user to install several external Python dependencies.
Packages: Includes pdfplumber, pandas, openpyxl, pytesseract, pdf2image, and pypdf.
Source: These are well-known, industry-standard libraries for data processing and PDF manipulation.
[COMMAND_EXECUTION]: The skill requires the execution of system-level commands for setup and operation.
Evidence: SKILL.md contains pip install commands for environment setup and notes a requirement for the tesseract binary to be installed on the host system OS.

pdf-to-structured